The present application relates to the field of artificial intelligence, and in particular to the technical fields of image processing, computer vision, deep learning, data processing, etc. Provided are an image processing method, a training method for an optical flow prediction model, and related apparatuses. The image processing method comprises: acquiring a portrait image to be processed; performing resolution scaling processing on said portrait image, so as to obtain a target-resolution image; on the basis of skeleton points and contour points of a target human body in the target-resolution image, acquiring a PAFs image corresponding to the target-resolution image; on the basis of the PAFs image and a skeleton map, performing optical flow prediction on the target-resolution image, so as to obtain a target optical flow map corresponding to the target-resolution image, wherein the skeleton map is obtained on the basis of the skeleton points; and on the basis of the target optical flow map, performing shaping processing on said portrait image, so as to obtain a shaped target image corresponding to said portrait image. The present application can perform shaping processing more accurately, and has a relatively good generalization performance.
The present disclosure relates to the field of artificial intelligence, in particular to the field of deep learning, and provides a large model-based video processing method and apparatus, a device, and a storage medium. A specific implementation solution is: extracting exercise information and knowledge information from a video to be processed; inputting the knowledge information into a preset lecture note large model for processing to generate content text of a lecture note; segmenting the content text of the lecture note to obtain a plurality of content blocks; and matching the exercise information with the content blocks, and inserting the exercise information into the content blocks matched therewith to obtain a file to be displayed of the lecture note.
G06F 16/70 - Recherche d’informationsStructures de bases de données à cet effetStructures de systèmes de fichiers à cet effet de données vidéo
G09B 5/08 - Matériel à but éducatif à commande électrique avec présentation individuelle d'une information à une pluralité de postes d'élèves
G06N 20/10 - Apprentissage automatique utilisant des méthodes à noyaux, p. ex. séparateurs à vaste marge [SVM]
H04N 21/431 - Génération d'interfaces visuellesRendu de contenu ou données additionnelles
H04N 21/44 - Traitement de flux élémentaires vidéo, p. ex. raccordement d'un clip vidéo récupéré d'un stockage local avec un flux vidéo en entrée ou rendu de scènes selon des graphes de scène du flux vidéo codé
3.
BARE METAL INSTANCE MANAGEMENT AND CONTROL METHOD AND APPARATUS, DEVICE, AND STORAGE MEDIUM
BAIDU.COM TIMES TECHNOLOGY (BEIJING) CO., LTD. (Chine)
Inventeur(s)
Miao, Kezhan
Ying, Ru
Zheng, Ran
Meng, Xianjun
Abrégé
The present disclosure relates to the technical field of computers, and in particular to the technical fields of cloud computing, cloud services, bare metal products and the like, and provides a bare metal instance management and control method and apparatus, a device, and a storage medium. The specific implementation solution is: a smart card receives an operation request for a bare metal instance; and on the basis of an interface provided by a virtualization library, the smart card performs, on the bare metal instance, a target operation corresponding to the operation request.
G06F 9/455 - ÉmulationInterprétationSimulation de logiciel, p. ex. virtualisation ou émulation des moteurs d’exécution d’applications ou de systèmes d’exploitation
4.
ANIMAL LANGUAGE CONVERSION METHOD AND APPARATUS, ANIMAL EMOTION RECOGNITION METHOD AND APPARATUS, AND MODEL TRAINING METHOD AND APPARATUS
The present disclosure relates to the technical field of artificial intelligence, and in particular to the technical fields of machine learning, deep learning, natural language processing, etc. Provided are an animal language conversion method and apparatus, an animal emotion recognition method and apparatus, and a model training method and apparatus. The specific implementation solution involves: acquiring multi-modal data related to an animal; preprocessing the multi-modal data to obtain fused multi-modal data; on the basis of the fused multi-modal data, recognizing the current emotion of the animal, so as to obtain an emotion recognition result for the animal; and performing semantic mapping and language translation on the emotion recognition result to convert an animal language into a human language, so as to obtain a language conversion result.
G10L 17/26 - Reconnaissance de caractéristiques spéciales de voix, p. ex. pour utilisation dans les détecteurs de mensongeReconnaissance des voix d’animaux
G10L 25/63 - Techniques d'analyse de la parole ou de la voix qui ne se limitent pas à un seul des groupes spécialement adaptées pour un usage particulier pour comparaison ou différentiation pour estimer un état émotionnel
G10L 15/24 - Reconnaissance de la parole utilisant des caractéristiques non acoustiques
G10L 25/30 - Techniques d'analyse de la parole ou de la voix qui ne se limitent pas à un seul des groupes caractérisées par la technique d’analyse utilisant des réseaux neuronaux
5.
LONG-TEXT GENERATION METHOD AND APPARATUS, AND DEVICE, STORAGE MEDIUM AND COMPUTER PROGRAM PRODUCT
The present disclosure relates to the technical field of big data processing, and in particular to the technical field of intelligent offices. Provided are a long-text generation method and apparatus, and an electronic device, a computer-readable storage medium and a computer program product. A specific embodiment of the method comprises: on the basis of long-text requirement information, generating a long-text outline, wherein the long-text outline comprises chapter entries; in response to file data associated with the chapter entries having been received, on the basis of the file data, generating text segments corresponding to the chapter entries; and on the basis of the long-text outline and the text segments corresponding to the chapter entries, generating long text.
The present disclosure provides an interleaved pipeline scheduling method and apparatus, a device, and a storage medium, relating to the technical field of computers, and in particular to the technical field of artificial intelligence such as deep learning, neural networks, and parallel computing. The specific implementation scheme comprises: on the basis of a pipeline segmentation dimension, an interleaving dimension, and an accumulation count, obtaining micro-batch scheduling parameters; on the basis of the micro-batch scheduling parameters, determining a transmission interval for micro-batches requiring buffered scheduling; and on the basis of the transmission interval and a buffer unit, performing buffered scheduling on computation results of micro-batches requiring delayed transmission. According to the present disclosure, a buffer unit, such as a buffer queue, can be created to buffer computation results from some computation units in the buffer unit, and to delay the transmission of the computation results. In this way, the transmission of computation results for some micro-batches can be delayed, thereby improving the applicability of interleaved pipeline scheduling.
A sparse processing method and apparatus for a sparse attention network, and an electronic device. The method specifically comprises: in a memory cell, using constraint conditions to construct a masked sparse representation (S301), wherein the constraint conditions are: the dimension of a mask matrix is [B,a,S], B represents the batch size, a represents the number of heads, S represents the sequence length, and each element in the S dimension represents a masked starting row of each column in the mask matrix; and a calculation unit acquiring the masked sparse representation from the memory cell, using the masked sparse representation to perform sparse processing on input data, and storing the sparse processing result into the memory cell (S302). According to the method, the constraint conditions are used to construct the masked sparse representation, so that memory consumption can be reduced from a quadratic order of the sequence length to a linear order of the sequence length, thereby remarkably reducing memory requirements during large model training, and improving the training efficiency.
The present disclosure relates to the technical field of computers, in particular to the technical fields such as data processing, difference detection, and software development, and provides a document difference detection method and apparatus, a device, and a storage medium. The specific implementation comprises: acquiring at least two documents to be detected, said at least two documents including documents at different stages in a collaborative design process; determining a plurality of pieces of content in each of said documents; determining binding relationships between content in different documents among said documents; and using a preset detection rule and the binding relationships to detect whether there is a difference between said at least two documents. According to the present disclosure, a difference between documents at different stages of collaborative design can be detected.
The present disclosure provides a text inference acceleration method applied to a large language model and a related device, relating to the technical field of data processing, and in particular to the technical fields of large language models, deep learning, and long text inference. The specific implementation scheme is as follows: selecting, from a target token set stored in a video memory, a core token set that needs to be retained, wherein the core token set at least comprises a first token subset, the first token subset is determined on the basis of attention scores obtained by performing a global query operation on the target token set using a plurality of proxy tokens, and each proxy token is selected from the target token set; and on the basis of the core token set, performing an eviction operation on the target token set in the video memory. In this embodiment of the present disclosure, a core token set that needs to be retained is selected by means of a plurality of proxy tokens, such that the processing burden caused by unnecessary tokens is reduced, thereby effectively freeing video memory space and improving the inference speed of the large language model.
An image annotation data generation method and apparatus, a model training method and apparatus, a device and a medium. The image annotation data generation method comprises: acquiring a sample image, a target recognition object in the sample image, and visual description information of the target recognition object (S101); verifying words in the visual description information on the basis of the sample image and the target recognition object to obtain a verification result of the visual description information (S102); adjusting the words in the visual description information on the basis of the verification result of the visual description information to obtain target description information of the target recognition object (S103); and taking the target description information of the target recognition object as annotation data of the target recognition object (S104).
An image labeling method and apparatus, a device, and a storage medium. The method comprises: using a preset labeling model to perform initial labeling processing on a target image to be labeled, to obtain target position information of a target object in the target image, and description information to be verified (S101); inputting the target image and target prompt information into a generative pre-trained transformer (GPT) model, to obtain an object search result, the target prompt information being used for prompting the GPT model to search, on the basis of the target description information, the target image for the target object and/or a virtual object associated with the target object, and the target description information being determined on the basis of the description information to be verified (S102); and when it is determined on the basis of the object search result that the verification of the description information to be verified is passed, determining labeling information of the target image on the basis of the description information to be verified and the target position information (S103).
The present application provides a method and apparatus for transmitting audio and video data, a device, and a medium. The method for transmitting audio and video data comprises: determining a plurality of candidate transmission sequences for data frames in audio and video data to be transmitted, the data frames comprising audio frames and video frames; on the basis of frame attributes of the data frames, determining a loss parameter of each candidate transmission sequence; and, on the basis of the loss parameters, determining a target transmission sequence from the plurality of candidate transmission sequences, and transmitting the data frames to a client by using the target transmission sequence.
H04N 21/262 - Ordonnancement de la distribution de contenus ou de données additionnelles, p. ex. envoi de données additionnelles en dehors des périodes de pointe, mise à jour de modules de logiciel, calcul de la fréquence de transmission de carrousel, retardement de la transmission de flux vidéo, génération de listes de reproduction
H04N 21/8547 - Création de contenu impliquant des marquages temporels pour synchroniser le contenu
13.
METHOD AND APPARATUS FOR CONFIGURING SCREEN, SYSTEM, ELECTRONIC DEVICE, AND STORAGE MEDIUM
The present disclosure provides a method and apparatus for configuring a screen, a system, an electronic device, a storage medium, and a program product, relating to the technical field of data processing, in particular to the technical field of display control and the technical field of large models. The specific implementation scheme comprises: on the basis of a received screen configuration statement, determining a display area to be configured and corresponding operation information; on the basis of layout information and the operation information of the display area, generating a screen configuration instruction; and, on the basis of the screen configuration instruction, determining display content and outputting the display content to the display area.
The present disclosure relates to the technical field of data processing, in particular to the technical fields such as map technology, indexing technology, trajectory positioning, and Internet of things technology, and provides a data processing method and apparatus, an electronic device and a computer-readable storage medium. The specific implementation solution comprises: acquiring index data, wherein the index data is obtained by clustering a plurality of area of interest spatial surfaces, the index data comprises a global index and a plurality of local indexes, the local indexes have one-to-one correspondence to clustering results and are relational indexes of one or more area of interest spatial surfaces comprised in the clustering results, and the global index is a relational index of the plurality of clustering results; and on the basis of the index data, acquiring from among full trajectory data the trajectory data corresponding to at least one area of interest spatial surface.
The present disclosure relates to the technical field of computers, and particularly relates to the technical fields of autonomous driving and artificial intelligence. Provided are an autonomous driving model based on temporal recursive autoregressive inference, and a method, an apparatus and a vehicle. In the autonomous driving model, a coding layer is configured to code sensor information of a current moment, so as to obtain a current scenario representation; a trajectory planning layer is configured to determine a driving trajectory from the current moment to a future moment on the basis of a historical scenario representation of the current moment; and an inference layer is configured to determine a predicted scenario representation of at least one future moment and a historical scenario representation of the future moment on the basis of the current scenario representation, the historical scenario representation and current prompt information, wherein the prompt information at least comprises the driving trajectory. Therefore, the autonomous driving model can use an integrated inference layer to realize the learning of historical information and the prediction of future information, so that the model can perform future prediction while learning the history, and thus the prediction effect of the model is improved.
The present disclosure relates to the technical field of computers, in particular to the technical field of autonomous driving and artificial intelligence, and provides a generative diffusion model-based autonomous driving model, a method, an apparatus, and a vehicle. An encoding layer in an autonomous driving model is configured to encode current perception information of an autonomous driving vehicle, to obtain a discrete spatial representation of a current scene, a prediction layer is configured to perform discrete diffusion on the basis of at least one scene discrete spatial representation comprising the discrete spatial representation of the current scene, to determine a prediction spatial representation at a future moment, and a decoding layer is configured to decode the prediction spatial representation, to obtain autonomous driving decision information at the future moment. Thus, according to the autonomous driving model, a generative diffusion model-based output can be used to determine the autonomous driving decision of the autonomous driving model. By improving the accuracy of future prediction, autonomous driving decision and prediction effects are further improved.
Provided are a map updating method and apparatus, a model training method and apparatus, an electronic device, and a medium, relating to the technical field of artificial intelligence, and in particular to the fields such as automatic driving, intelligent traffic, computer vision, and image processing. A specific implementation solution is: for a same piece of position information, determining associated historical map data and M pieces of collected image data based on a time sequence, wherein M is an integer greater than or equal to 1; respectively encoding the historical map data and the M pieces of collected image data to obtain a historical map feature and M collected image features; determining position information and category information of a target instance on the basis of the historical map feature and the M collected image features, wherein the target instance represents lane information; and updating the map data on the basis of the position information and the category information of the target instance.
G06F 16/29 - Bases de données d’informations géographiques
G06F 16/56 - Recherche d’informationsStructures de bases de données à cet effetStructures de systèmes de fichiers à cet effet de données d’images fixes en format vectoriel
G06F 16/587 - Recherche caractérisée par l’utilisation de métadonnées, p. ex. de métadonnées ne provenant pas du contenu ou de métadonnées générées manuellement utilisant des informations géographiques ou spatiales, p. ex. la localisation
G06F 18/214 - Génération de motifs d'entraînementProcédés de Bootstrapping, p. ex. ”bagging” ou ”boosting”
G06V 20/56 - Contexte ou environnement de l’image à l’extérieur d’un véhicule à partir de capteurs embarqués
G06V 20/70 - Étiquetage du contenu de scène, p. ex. en tirant des représentations syntaxiques ou sémantiques
G06V 10/26 - Segmentation de formes dans le champ d’imageDécoupage ou fusion d’éléments d’image visant à établir la région de motif, p. ex. techniques de regroupementDétection d’occlusion
G06V 10/80 - Fusion, c.-à-d. combinaison des données de diverses sources au niveau du capteur, du prétraitement, de l’extraction des caractéristiques ou de la classification
G01C 21/00 - NavigationInstruments de navigation non prévus dans les groupes
18.
AUTONOMOUS DRIVING MODEL, METHOD, APPARATUS AND VEHICLE CAPABLE OF ACHIEVING MULTI-MODAL INTERACTION
Provided are an autonomous driving model, method, apparatus and vehicle capable of achieving multi-modal interaction. An input layer (210) in an autonomous driving model (200) is configured to receive historical decision information (201), sensing information (202), traffic information (203) and interaction information (204) at a current moment; an encoding layer (220) is configured to encode input information; an autoregressive inference layer (230) is configured to obtain a hidden state for a next moment; and a decoding layer (240) is configured to perform decoding so as to obtain interaction information for the next moment and autonomous driving decision-making information for the next moment. The autonomous driving model (200) may understand a current driving environment by inferring the sensing information (202), the traffic information (203) and the interaction information (204), and gains a better understanding of the influence of historical operations on the autonomous driving process by inferring historical decision-making data, thereby making an output result of the autonomous driving model achieve better interpretability and controllability.
The present disclosure relates to the field of computer technology and particularly relates to the technical fields of autonomous driving and artificial intelligence. Provided are an autonomous driving method, apparatus and vehicle capable of following instructions for self-recovery. The autonomous driving method comprises: acquiring input information; encoding the input information to obtain an input tensor corresponding to the input information; performing autoregressive inference on the input tensor to obtain a hidden state for a first moment after a current moment; and performing decoding on the basis of the hidden state of the first moment to obtain interaction information for the first moment and autonomous driving decision-making information for the first moment, wherein the interaction information for the first moment comprises a signal used for indicating that assistance is required during autonomous driving. In this way, prompt information for natural language interaction can be used to instruct an autonomous driving model to control a vehicle, thereby achieving the rapid generation of a self-recovery scheme. In addition, the vehicle can perform autonomous self-recovery on the basis of self-recovery scheme instructions.
B60W 60/00 - Systèmes d’aide à la conduite spécialement adaptés aux véhicules routiers autonomes
G05D 1/00 - Commande de la position, du cap, de l'altitude ou de l'attitude des véhicules terrestres, aquatiques, aériens ou spatiaux, p. ex. utilisant des pilotes automatiques
20.
METHOD AND APPARATUS FOR DETECTING FAILURE TO YIELD TO PEDESTRIANS OF VEHICLE
The present disclosure relates to the technical field of artificial intelligence such as intelligent transportation and computer vision, and provides a method and apparatus for detecting failure to yield to pedestrians of a vehicle, an electronic device, and a readable storage medium. The method for detecting failure to yield to pedestrians of a vehicle comprises: acquiring a marked crosswalk area on the basis of surveillance images in a surveillance video stream; for each surveillance image frame in the surveillance video stream, acquiring the intersection over union between a target vehicle and the marked crosswalk area in the surveillance image frame, and selecting multiple surveillance image frames having intersection over union greater than a preset intersection over union threshold as target surveillance images corresponding to the target vehicle; and sequentially performing human detection on each target surveillance image frame in descending order of intersection over union, and when it is determined that a human detection result corresponding to a current target surveillance image satisfies a preset requirement, determining that the target vehicle is a regulation-violating vehicle which does not yield to pedestrians. The present disclosure can expand the detection scenario, reduce the computing resources required for detection, and improve the detection efficiency while detecting whether vehicles yield to pedestrians.
The present disclosure relates to the technical fields of artificial intelligence, such as cloud services, big data and large language models, and provides an algorithm service deployment method and apparatus, an electronic device, and a readable storage medium. The algorithm service deployment method comprises: a first server acquiring service operation data and service information of a target algorithm service, wherein the service operation data comprises one of an application program interface, a mirror image file and a model file; determining a service access type on the basis of the service property of the target algorithm service, and acquiring a target access method corresponding to the service access type, wherein the service access type comprises one of application program interface access, mirror image access and model access; and deploying the target algorithm service at the first server on the basis of the target access method, the service operation data and the service information. According to the present disclosure, the first server supports deployment of algorithm services corresponding to different service access types, so that the first server can have stronger service deployment performance.
The present disclosure relates to the technical field of artificial intelligence, in particular to the technical fields of computer vision, deep learning and the like, and provides a target identification method and device. The specific implementation comprises: obtaining a video stream collected by a rotatable camera device; determining a preset position of the camera device on the basis of a target in an image frame corresponding to the video stream; determining a still image frame and a fixed still object in the still image frame on the basis of the preset position and the video stream; and determining a position change result of the target in the video stream on the basis of the still image frame and the fixed still object. The implementation improves the accuracy of position change identification of targets.
The present disclosure relates to the technical field of artificial intelligence, and specifically to technical fields such as computer vision, deep learning and big data. Provided are a rainfall identification method and apparatus, a model training method and apparatus, and a device and a storage medium, which may be applied in scenarios such as smart cities and emergency management. The rainfall identification method comprises: processing a target video collected by a target camera, so as to obtain an initial rainfall identification result; determining a target rainfall amount station, the distance between which and the target camera meets a preset condition, and acquiring target rainfall amount data of the target rainfall amount station; and on the basis of the initial rainfall identification result and the target rainfall amount data, determining a target rainfall identification result. The present disclosure may improve the accuracy of rainfall identification.
Provided are a human-computer interaction method and apparatus, an electronic device and a storage medium, relating to the technical field of artificial intelligence, in particular to the technical fields of deep learning, natural language processing and large models. The specific implementation solution comprises: in response to a human-computer interaction request and on the basis of a first dialogue text comprised in the human-computer interaction request, determining from among a plurality of plug-ins registered in a large language model a first target plug-in related to the first dialogue text; obtaining a second dialogue text on the basis of the first dialogue text and a description text of the first target plug-in; and inputting the second dialogue text into the large language model to obtain a reply text.
The present disclosure relates to the technical fields of computers, and in particular to the technical fields of artificial intelligence (AI), neural network models, and smart city, and provides a target detection method based on a multi-task AI large model, and a model training method based on a multi-task AI large model. The specific implementation scheme of the target detection method comprises: recognizing a target object of an image under test to obtain a first recognition result; on the basis of the confidence level of the first recognition result and a first threshold corresponding to a first precision, determining a first alarm object from the first recognition result as a detection result; when a trigger condition is met, performing target detection on an image under supplementary test corresponding to the first recognition result to obtain a second recognition result; on the basis of the confidence level of the second recognition result and a second threshold corresponding to a second precision, determining a second alarm object from the second recognition result; and updating the detection result on the basis of the second alarm object. The present disclosure can ensure high precision of target detection and reduce the missing recall.
The present application relates to the technical field of computers, and particularly relates to the field of source codes. Provided are a method and apparatus for generating simulated data, and an electronic device and a storage medium. The specific implementation solution comprises: acquiring a database table of a project to be processed, and determining data simulation configuration information corresponding to a field in the database table, wherein the database table comprises at least one field, which represents the type of simulated data required by said project, and the data simulation configuration information represents a generation means for the simulated data and the format of the simulated data; and on the basis of the data simulation configuration information, generating the simulated data under the field in the database table. Different pieces of configuration information are set for different fields, and when it is necessary to simulate data, corresponding configuration information is searched for to automatically generate simulated data, thereby improving the simulation efficiency of the data.
Disclosed are a sentiment analysis method and apparatus, a large language model training method and apparatus, an electronic device, a storage medium, a computer program product and a computer program. The sentiment analysis method comprises: acquiring first target text; extracting from the first target text an object to be analyzed; generating second target text on the basis of the first target text and said object, wherein the second target text comprises task prompt text, and the task prompt text is used for prompting a large language model to execute a sentiment analysis task on said object on the basis of the first target text; and inputting the second target text into the large language model to obtain the sentiment polarity of said object.
The present application provides a large language model-based event processing method and apparatus, a device and a medium. The large language model-based event processing method comprises: acquiring a question input by a user; acquiring pre-generated event information; fusing the question and the event information to obtain a fusion result; and inputting the fusion result into a pre-trained large language model to obtain reply content corresponding to the question.
The present disclosure relates to the technical field of artificial intelligence, and specifically to the technical fields such as large models and natural language understanding, and provides a chart generation method and apparatus, a device, and a storage medium. The chart generation method comprises: acquiring target text content and target prompt information; on the basis of the target text content and the target prompt information, using a first pre-trained language model to generate structured information, wherein the structured information is used for generating a target chart; on the basis of the structured information, using a second pre-trained language model to generate the target chart; and displaying the target chart. The present disclosure can improve the chart generation efficiency and accuracy.
The present disclosure relates to the technical field of video processing, in particular to the technical field of monitoring video processing, and provides a monitoring video processing method, a monitoring video processing apparatus, an electronic device, a storage medium, and a program product. The specific implementation solution is: acquiring a monitoring video stream to be processed; performing semantic segmentation on video frames in said monitoring video stream to obtain semantic tags of the video frames; on the basis of the semantic tags of the video frames and scenario determination rules, determining service scenarios to which the video frames are applicable; and determining a scenario tag of said monitoring video stream on the basis of the service scenarios to which the plurality of video frames in said monitoring video stream are applicable.
The present disclosure relates to the technical field of code generation, and provides a code generation method and apparatus, a device and a storage medium. The method comprises: in response to receiving operation information of a user, requesting a node backend to create a corresponding project and a project file; in response to determining that the project and the project file are created, requesting a java backend to create a data source; calling the java backend to perform project engineering initialization and code assembly; and in response to determining that the code assembly is completed, performing project verification and code export.
Disclosed are a student model generation method and apparatus based on a large model, and an electronic device, a storage medium, a computer program product and a computer program. The method comprises: acquiring a sample data set; inputting input data and prompt information into a large model, so as to acquire first content generated by the large model; converting the first content into a first prediction result which has the same type as a labeling result; inputting the input data into an initial student model, so as to acquire a second prediction result that is output by the initial student model; determining a correction gradient on the basis of respective differences between the second prediction result and the first prediction result and between the second prediction result and the labeling result; and on the basis of the correction gradient, correcting the initial student model, so as to acquire a target student model.
The present disclosure relates to the technical field of artificial intelligence, and specifically relates to the fields of deep learning and computer vision. Provided are a multi-objective optimization method and apparatus, and a device and a storage medium. The method comprises: acquiring a set of values to be quantized and a set of objectives to be optimized; for each value to be quantized in said set of values, acquiring a quantization coefficient corresponding to said value, and determining a set of adjacent quantization coefficients of the quantization coefficient; on the basis of a set of reconstruction values corresponding to the set of adjacent quantization coefficients, determining a set of reconstruction distortion values; and on the basis of the set of reconstruction distortion values and said set of objectives, determining a target quantization coefficient. The multi-objective optimization method provided in the present disclosure improves the performance of other optimization objectives while ensuring the performance of traditional optimization objectives.
G06V 10/74 - Appariement de motifs d’image ou de vidéoMesures de proximité dans les espaces de caractéristiques
G06V 10/762 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant le regroupement, p. ex. de visages similaires sur les réseaux sociaux
METHOD AND APPARATUS FOR GENERATING IMAGE ADVERSARIAL SAMPLE, METHOD AND APPARATUS FOR TRAINING IMAGE PROCESSING MODEL, IMAGE PROCESSING METHOD AND APPARATUS, AND DEVICE AND MEDIUM
Provided in the present application are a method and apparatus for generating an image adversarial sample, a method and apparatus for training an image processing model, an image processing method and apparatus, and a device and a medium. The method for generating an image adversarial sample comprises: acquiring an original image sample, and acquiring a feature vector map corresponding to the original image sample (S110); performing image scaling processing on the feature vector map according to an image scale of the original image sample, so as to obtain a standard-scale feature map (S120); using an attention mechanism network of a target type to process the standard-scale feature map, so as to obtain an attention influence map (S130); and on the basis of the attention influence map, adding a disturbance to the original image sample, so as to obtain an image adversarial sample corresponding to the original image sample (S140).
G06V 10/774 - Génération d'ensembles de motifs de formationTraitement des caractéristiques d’images ou de vidéos dans les espaces de caractéristiquesDispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant l’intégration et la réduction de données, p. ex. analyse en composantes principales [PCA] ou analyse en composantes indépendantes [ ICA] ou cartes auto-organisatrices [SOM]Séparation aveugle de source méthodes de Bootstrap, p. ex. "bagging” ou “boosting”
35.
LARGE LANGUAGE MODEL-BASED DATA QUERY METHOD AND APPARATUS, DEVICE, AND STORAGE MEDIUM
The present disclosure relates to the technical field of artificial intelligence, in particular to the technical fields such as intelligent office, cloud computing, generative dialogue systems, and large language models (LLMs), and provides an LLM-based data query method and apparatus, a device, and a storage medium. The LLM-based data query method comprises: determining a target data table from among candidate data tables on the basis of a query question, wherein the target data table comprises candidate attributes; determining a target attribute from among the candidate attributes on the basis of the query question; generating query instruction prompt information of an LLM on the basis of the query question, table information of the target data table, and attribute information of the target attribute, and using the LLM to generate a query instruction on the basis of the query instruction prompt information; and on the basis of the query instruction, querying from within the target data table to obtain a query answer corresponding to the query question. The present disclosure can improve the data query efficiency and accuracy.
The present disclosure relates to the technical field of artificial intelligence, and in particular relates to the technical fields of deep learning, natural language processing, etc. Provided are a model compression method and apparatus, a training method and apparatus, and a text data processing method and apparatus. A specific implementation of the model compression method involves: according to the number of concurrently deployed calculation units, dividing initial model parameters of a model to be compressed, so as to obtain initial local model parameters corresponding to the plurality of calculation units, wherein the calculation units are used for processing the same task associated with the model to be compressed; according to an initial input activation value, rematching the correspondence between initial weight parameters and the calculation units, so as to obtain target local model parameters corresponding to the plurality of calculation units, wherein a target input activation value among the target local model parameters gradually increases in a data processing direction; and performing quantization on the target local model parameters, so as to obtain a compressed model.
The present application provides a compound for inhibiting protein-protein interaction between CDK4/6 and cyclin D. The compound has the result shown in formula (I), formula (II), formula (III), formula (IV), formula (V) or formula (VI), and has a use in preparation of drugs for diseases that benefit from or can be treated by inhibiting protein-protein interaction between CDK4/6 and cyclin D.
C07J 9/00 - Stéroïdes normaux contenant du carbone, de l'hydrogène, un halogène ou de l'oxygène, substitués en position 17bèta par une chaîne de plus de deux atomes de carbone, p. ex. cholane, cholestane, coprostane
C07D 403/02 - Composés hétérocycliques contenant plusieurs hétérocycles, comportant des atomes d'azote comme uniques hétéro-atomes du cycle, non prévus par le groupe contenant deux hétérocycles
C07D 413/00 - Composés hétérocycliques contenant plusieurs hétérocycles, au moins un cycle comportant des atomes d'azote et d'oxygène comme uniques hétéro-atomes du cycle
In one embodiment, a method for controlling PCIe devices on an autonomous driving system (ADS) of an autonomous driving vehicle (ADV) is disclosed. The method includes scanning PCIe ports of the ADS to discover PCIe device (s) mounted on the ADS. Next, a comparison is performed between a list of the discovered PCIe device (s) and an expected list of PCIe devices for the ADS. Next, an offline PCIe device is determined based on the comparison, the offline PCIe device corresponding to a PCIe device that is present in the expected list of PCIe devices but not in the list of the discovered PCIe devices. Then a device reset command is transmitted to a programmable logic device that manages power for PCIe devices of the ADS to reset the offline PCIe device, wherein the programmable logic device generates a reset control signal for the offline PCIe device to reset the PCIe device.
A printed circuit board (PCB) includes one or more signal layers in between a first surface and a second surface, the one or more signal layers including a first signal layer. The PCB includes a connector interface disposed on the first surface and a first signal via electrically coupled to the connector interface. The PCB includes first transmission line disposed at the first signal layer electrically coupling a first location to a second location of the first signal layer and a second signal via disposed at the second location. The PCB includes a power-over-cable circuit disposed on the first surface and electrically coupled to the second signal via and a receiver circuit disposed on the first surface adjacent to the power-over-cable circuit. The PCB includes a second transmission line disposed on the first surface layer electrically coupling the power-over-cable circuit to the receiver circuit.
A system (320) monitors a temperature of a computing system for an autonomous driving system (ADS) (110) of the autonomous driving vehicle (ADV) (101). The system (320) determines the monitored temperature is below a predetermined temperature threshold. The system (320) controls a heating module (407) to distribute heat to the computing system. The system (320) determines the temperature of the computing system is above a predetermined temperature threshold. The system (320) initiates a boot up sequence to boot up the computing system, where the temperature of components of the computing system is above the predetermined threshold when the boot up sequence is initiated.
In one embodiment, a microcontroller unit (MCU) receives an expected state of an autonomous driving vehicle (ADV) from a controller of the ADV(101), where the controller controls motions of the ADV (101)using a control algorithm. The MCU(321) eceives sensor data from one or more sensors of the ADV(101). The MCU (321)determine an actual state of the ADV (101)based on the sensor data. The MCU (321)determines a performance metric(505) of the control algorithm based on the expected state and the actual state. In response to determining the performance metric has satisfied a predetermined condition, the MCU(321) determines a plurality of weight values for the control algorithm. The MCU(321) sends the plurality of weight values to the control system to tune one or more weight parameters of the control algorithm using the plurality of weight values, where the controller controls the ADV(101) using the tuned control algorithm.
B60W 40/02 - Calcul ou estimation des paramètres de fonctionnement pour les systèmes d'aide à la conduite de véhicules routiers qui ne sont pas liés à la commande d'un sous-ensemble particulier liés aux conditions ambiantes
42.
ARTIFICIAL INTELLIGENCE ENABLED VEHICLE OPERATING SYSTEM
A vehicle operating system (VOS) in an autonomous driving vehicle (ADV) can communicate with a cloud platform to automatically train AI models. The VOS collects real-time data from the ADV, and generates inference data based on the real-time data using a teacher edge model of an AI model and generates second inference data based on the real-time data using a student edge model of the AI model (601). The VOS then obtains one or more differences between the first inference data and the second inference data (603), and retrains the student edge model of the Al model based on the one or more differences (605). Both real-time data and the retrained student edge model are uploaded to a cloud platform for use in upgrading the student edge model and the teacher edge model on the cloud platform. The upgraded teacher edge model and the student edge model can be redeployed over-the-air (OTA) through a software define process. The above process of training AI models can be repeated in a closed-loop automatically without user intervention.
G05B 19/042 - Commande à programme autre que la commande numérique, c.-à-d. dans des automatismes à séquence ou dans des automates à logique utilisant des processeurs numériques
G06F 9/451 - Dispositions d’exécution pour interfaces utilisateur
Provided in the present disclosure are a multi-modal-fused method and apparatus for recognizing a high-definition map element, and a device and a medium. The multi-modal-fused method for recognizing a high-definition map element comprises: on the basis of point cloud data of a target map element and at least two types of candidate image data, determining attribute features of the target map element, a pixel registration feature thereof and a hybrid registration feature thereof; according to the pixel registration feature of the target map element, determining a pixel correspondence of the target map element, and according to the hybrid registration feature of the target map element, determining a hybrid correspondence of the target map element; and on the basis of the pixel correspondence and the hybrid correspondence, performing fusion processing on the attribute features of the target map element, so as to obtain a fused feature of the target map element, and on the basis of the fused feature of the target map element, determining an element category of the target map element.
G06V 10/80 - Fusion, c.-à-d. combinaison des données de diverses sources au niveau du capteur, du prétraitement, de l’extraction des caractéristiques ou de la classification
44.
SPEECH RECOGNITION METHOD AND APPARATUS, AND DEVICE AND MEDIUM
A speech recognition method and apparatus, and a device and a medium. The method comprises: using an acoustic model to process speech data to be recognized and a first text segment that has been recognized, so as to obtain respective acoustic probabilities of a plurality of alternative text segments (S210); using a first language sub-model to process the first text segment, so as to obtain respective initial language probabilities of the plurality of alternative text segments (S220); using a constraint sub-model to process the first text segment, so as to obtain respective extensible relationships of the plurality of alternative text segments with respect to the first text segment (S230); adjusting the initial language probabilities of the alternative text segments according to the extensible relationships, so as to obtain respective first language probabilities of the plurality of alternative text segments (S240); and determining a target text segment from among the plurality of alternative text segments according to the first language probabilities and the acoustic probabilities (S250).
The present disclosure relates to the technical field of artificial intelligence, in particular, to the fields of deep learning, user coding/compiling and the like. Provided are a model debugging method and apparatus, an electronic device, and a storage medium. The model debugging method comprises: loading a debugging tool in a dynamic graph mode; matching operator composition information corresponding to at least one component in a network model with an operator fusion policy in the debugging tool to obtain a matching result; and according to the matching result, obtaining debugging information for model debugging.
The present disclosure provides a multi-task model training method, relating to the technical field of artificial intelligence, and in particular, to the technical fields of deep learning, cloud computing, multi-task parallel processing, and data searching. A specific implementation solution comprises: inputting sample behavior data of a sample object into a shared sub-model, to obtain behavior characteristic information of the sample object, the behavior characteristic information comprising multiple pieces of behavior characteristic sub-information; obtaining a sub-loss value of a task processing sub-model according to the behavior characteristic sub-information associated with the task processing sub-model; determining a target gradient value corresponding to the sub-loss value, according to the sub-loss value of the task processing sub-model; determining a weight value corresponding to the sub-loss value, according to multiple target gradient values; and training a multi-task model according to multiple sub-loss values and the multiple weight values corresponding to the multiple sub-loss values. The present disclosure further provides an information recommendation method, an apparatus, an electronic device, and a storage medium.
The present invention relates to the technical field of artificial intelligence, and provides text image generation, training, and processing methods, and an electronic device. A specific implementation solution comprises: dividing a sample text image set into at least one sample text image subset, according to a sample text output result set and a sample label set of the sample text image set (S210); according to a sample text output result set of a set of sample text images to be clipped, determining a target clipping location set of the set of sample text images to be clipped (S220); clipping the set of sample text images to be clipped on the basis of the target clipping location set, to obtain at least one clipped sample text image subset (S230); obtaining a target sample text image set according to the at least one clipped sample text image subset and the at least one sample text image subset (S240). The accuracy of the target clipping location can be effectively ensured, character information is effectively prevented from being damaged, and image background complexity and image diversity of sample text images in the target sample text image set are improved.
G06V 10/774 - Génération d'ensembles de motifs de formationTraitement des caractéristiques d’images ou de vidéos dans les espaces de caractéristiquesDispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant l’intégration et la réduction de données, p. ex. analyse en composantes principales [PCA] ou analyse en composantes indépendantes [ ICA] ou cartes auto-organisatrices [SOM]Séparation aveugle de source méthodes de Bootstrap, p. ex. "bagging” ou “boosting”
48.
INFORMATION INTERACTION METHOD AND APPARATUS, DEVICE AND MEDIUM
The present disclosure relates to the technical field of artificial intelligence, in particular to the technical fields of natural language processing and deep learning. Provided are an information interaction method and apparatus, a device and a medium. The implementation solution comprises: acquiring first dialogue data of a user; processing the first dialogue data by means of a dialogue system to acquire first reply data corresponding to the first dialogue data; outputting the first reply data; in response to failing to receive, within a preset time after the first reply data is output, second dialogue data sent by the user, processing the first reply data by means of the dialogue system to acquire second reply data; and outputting the second reply data.
Provided are an image processing method and apparatus, and an electronic device, a storage medium, a computer program product and a computer program. The image processing method comprises: acquiring an image to be processed; and inputting said image into a target image model, such that the target image model outputs a processing result of said image.
The present disclosure relates to the technical field of computers, and in particular to the technical field of virtual keyboards. Provided are a virtual keyboard presentation method and apparatus, and a storage medium, an electronic device and a product. The specific implementation solution involves: acquiring an interface display mode of a virtual keyboard; acquiring information corresponding to the interface display mode; and presenting the interface display mode on the virtual keyboard on the basis of the information. By means of the present disclosure, the display of a virtual keyboard can be more intelligent, and the experience and feeling of intelligence of a user with respect to the virtual keyboard can be enhanced.
G06F 3/04886 - Techniques d’interaction fondées sur les interfaces utilisateur graphiques [GUI] utilisant des caractéristiques spécifiques fournies par le périphérique d’entrée, p. ex. des fonctions commandées par la rotation d’une souris à deux capteurs, ou par la nature du périphérique d’entrée, p. ex. des gestes en fonction de la pression exercée enregistrée par une tablette numérique utilisant un écran tactile ou une tablette numérique, p. ex. entrée de commandes par des tracés gestuels par partition en zones à commande indépendante de la surface d’affichage de l’écran tactile ou de la tablette numérique, p. ex. claviers virtuels ou menus
51.
TARGET SORTING MODEL TRAINING METHOD AND APPARATUS, TARGET SORTING METHOD AND APPARATUS, ELECTRONIC DEVICE AND STORAGE MEDIUM
Provided are a target sorting model training method and apparatus, a target sorting method and apparatus, an electronic device, a storage medium and a computer program product, which relate to the technical field of computers, in particular to the fields of data processing, machine learning, intelligent recommendation and so on. The target sorting model training method comprises: acquiring features respectively corresponding to N targets; by means of a shared expert sub-network in a target sorting model, extracting a common feature of the N targets from the features respectively corresponding to the N targets; by means of expert sub-networks respectively corresponding to the N targets in the target sorting model, extracting individual features respectively corresponding to the N targets from the features respectively corresponding to the N targets, wherein M targets amongst the N targets correspond to the same expert sub-network, M is an integer greater than or equal to 1, and M is less than N; and on the basis of the individual features respectively corresponding to the N targets and the common feature of the N targets, adjusting the parameters of the target sorting model.
An intelligent voice assistant continuous learning method and apparatus, an electronic device, and a storage medium. The method comprises: acquiring user voice input information (S101); acquiring user feedback information after the intelligent voice assistant has responded on the basis of the user voice input information (S102); based on the user feedback information, controlling the intelligent voice assistant to perform model optimization and updating (S103). The method effectively improves the continuous learning capabilities of an intelligent voice assistant, and optimizes the performance of the intelligent voice assistant.
A voice wakeup method and apparatus, an electronic device, a storage medium, and a program product. The method comprises: performing word recognition on a speech to be recognized, so as to obtain a wakeup word recognition result (S210); when it is determined that the wakeup word recognition result is used for representing that said speech comprises a predetermined wakeup word, performing syllable recognition on said speech to obtain a wakeup syllable recognition result (S220); and when it is determined that the wakeup syllable recognition result is used for representing that said speech comprises a predetermined syllable, determining that said speech is a correct wakeup speech (S230).
G10L 15/06 - Création de gabarits de référenceEntraînement des systèmes de reconnaissance de la parole, p. ex. adaptation aux caractéristiques de la voix du locuteur
G10L 15/20 - Techniques de reconnaissance de la parole spécialement adaptées de par leur robustesse contre les perturbations environnantes, p. ex. en milieu bruyant ou reconnaissance de la parole émise dans une situation de stress
54.
SPEECH RECOGNITION MODEL TRAINING METHOD AND APPARATUS, ELECTRONIC DEVICE, AND STORAGE MEDIUM
A speech recognition model training method and apparatus, an electronic device (1200), and a storage medium, relating to the technical field of artificial intelligence, and in particular to the fields of deep learning, speech recognition and the like. The method comprises: S201, constructing a negative sample according to a positive sample, so as to obtain a target negative sample used for constraining a speech decoding path; S202, obtaining training data according to the positive sample and the target negative sample; and S203, training a first speech recognition model according to the training data, so as to obtain a second speech recognition model.
G10L 15/06 - Création de gabarits de référenceEntraînement des systèmes de reconnaissance de la parole, p. ex. adaptation aux caractéristiques de la voix du locuteur
55.
SIGNAL MANAGEMENT AND CONTROL METHOD BASED ON VEHICLE INFRASTRUCTURE COOPERATION, AND RELATED APPARATUS AND PROGRAM PRODUCT
A signal management and control method based on vehicle infrastructure cooperation, and a related apparatus and a program product, which relate to the technical fields of vehicle infrastructure cooperation, intelligent agents, signal management and control, etc., and can be applied to intelligent traffic scenarios. The process (200) of the signal management and control method comprises: respectively taking traffic lights in a management and control network, intelligent vehicles, and map applications in a navigation mode as first intelligent agents, second intelligent agents, and third intelligent agents, wherein for an intelligent agent of each type, an action to be executed and corresponding to the type to which the intelligent agent belongs is preset (201); acquiring the current travel state parameter set in the management and control network (202); on the basis of the idea of stochastic gaming and a preset reward function, which minimizes the total travel duration of all people and all vehicles in the management and control network, determining, from among actions to be executed and corresponding to the intelligent agents, target actions to be executed and corresponding to the travel state parameter set (203); and issuing the target actions to be executed of the intelligent agents to the corresponding intelligent agents (204). Therefore, the total travel duration of objects in the management and control network can be the shortest.
The present disclosure relates to the field of computers, in particular to the technical field of intelligent recommendation, and provides an information recommendation method and apparatus, an electronic device, a computer readable storage medium, and a computer program product. An implementation solution is: obtaining browsed information lists of a plurality of first users and a first vector corresponding to each browsed information list; clustering the first vectors corresponding to the plurality of first users, so as to obtain one or more vector clusters and center vectors thereof; determining one or more information clusters respectively corresponding to the one or more vector clusters; in response to a browsing request of a second user, obtaining a browsed information list of the second user; in response to determining that the browsed information list of the second user is not empty, determining a second vector corresponding to the browsed information list of the second user; performing similarity calculation on the second vector and the center vectors, respectively, so as to determine an information cluster matching the second vector; and recommending the second user on the basis of the determined information cluster.
The present disclosure relates to the technical field of computers, and in particular to the technical field of artificial intelligence. Provided are a resource recognition method and apparatus, and a device and a storage medium. The specific implementation scheme involves: acquiring posterior information and prior information of a resource to be recognized, wherein the posterior information is used for reflecting feedback information of a user for said resource, and the prior information is used for reflecting semantic information of said resource; recognizing said resource according to a first recognition model and the posterior information, so as to obtain a first recognition result; recognizing said resource according to a second recognition model and the prior information, so as to obtain a second recognition result; and generating a third recognition result according to the first recognition result and the second recognition result. By means of the resource recognition method and apparatus, and the device and the storage medium provided in the present disclosure, the accuracy of resource recognition can be improved.
A method and apparatus for generating a digital person, a method and apparatus for training a model, and a device and a medium, which relate to the field of artificial intelligence, and in particular to the technical fields of natural language processing, deep learning, computer vision, image processing, augmented reality, virtual reality, etc., and can be applied to scenarios such as metaverse. The method for generating a digital person comprises: acquiring material content (S201); on the basis of a pre-trained scenario classification model, determining a plurality of scenarios from the material content (S202), wherein each scenario among the plurality of scenarios corresponds to one content fragment with complete semantic information in the material content; for each scenario among the plurality of scenarios, on the basis of the corresponding content fragment, determining target content corresponding to the scenario (S203); on the basis of the corresponding target content, determining scenario label information of the scenario (S204); and on the basis of the scenario label information, configuring a digital person specific to the scenario (S205).
A method for navigating while walking at night (200). The method comprises: determining a walking navigation start point and a walking navigation end point according to an obtained night walk travel request (201); determining candidate walking routes according to the walking navigation start point and the walking navigation end point (202); determining a safe night walk route according to the degrees of safety of walking at night on the candidate walking routes (203); and outputting night walk navigation content corresponding to the safe night walk route (204). According to the described method, by determining the degree of safety of walking at night on each candidate walking route, that is, by considering the impact on the safety of walking due to it being night, a candidate walking route that can be safely walked at night is further determined as a safe night walk route, so that navigation content having a relatively high safety can be provided to a user, thus the safety of a user walking at night and his or her property is guaranteed. Also provided are an apparatus for navigating while walking at night (900), an electronic device (1000), a computer-readable storage medium, and a computer program product.
Disclosed are an end-to-end sensitive text recall model training method and a sensitive text recall method. The specific implementation solution comprises: obtaining a preset word list and a first random text corpus in a sensitive text interception scene; constructing positive sample data according to the preset word list, and constructing negative sample data according to the first random text corpus; according to the positive sample data and the negative sample data, performing cyclic iterative training on an initial text classification model by using a manual evaluation mode and a multi-sample splicing sampling mode to obtain a text classification model a model index of which reaches a target index after training is completed; and generating an end-to-end sensitive text recall model according to model parameters of the text classification model the model index of which reaches the target index.
A task processing method implemented by using an integer programming solver, a device, and a medium, relating to the technical field of computers, in particular to artificial intelligence. The implementation solution comprises: acquiring distribution information of at least one variable that is not currently fixed among a plurality of variables related to a task (S201), the distribution information comprising type distribution information of the at least one variable and type distribution information of the at least one variable in an objective function of the task; on the basis of the distribution information of the at least one variable, determining a first target variable selection policy of a current question from among a plurality of candidate variable selection policies (S202); and determining at least one target variable of the current question from among the plurality of variables by using the first target variable selection policy (S203).
A sorting method for search results. The method comprises: acquiring a plurality of first search results corresponding to a search request and an initial order of the plurality of first search results; when at least one first search result is push information, determining the type and style of the at least one first search result; on the basis of a preset mapping relationship and according to the type and style of each first search result in the at least one first search result, determining a first score corresponding to each first search result; and adjusting the initial order according to a descending order of the first scores, so as to acquire an updated order of the plurality of first search results.
G06F 16/2457 - Traitement des requêtes avec adaptation aux besoins de l’utilisateur
G06F 16/215 - Amélioration de la qualité des donnéesNettoyage des données, p. ex. déduplication, suppression des entrées non valides ou correction des erreurs typographiques
G06N 3/04 - Architecture, p. ex. topologie d'interconnexion
63.
IMAGE SEGMENTATION SAMPLE GENERATION METHOD AND APPARATUS, METHOD AND APPARATUS FOR PRE-TRAINING IMAGE SEGMENTATION MODEL, AND DEVICE AND MEDIUM
Provided in the present disclosure are an image segmentation sample generation method and apparatus, a method and apparatus for pre-training an image segmentation model, and a device and a medium. The image segmentation sample generation method comprises: acquiring an image classification sample, wherein the image classification sample comprises a sample image and a classification label of the sample image; determining, by means of an interpretable algorithm, a weight value of each pixel point of the sample image under the action of an image classification model; selecting a forward label pixel point from the sample image according to the weight value of each pixel point; and forming, according to the forward label pixel point and the classification label of the sample image, an image segmentation sample corresponding to the image classification sample.
G06V 10/774 - Génération d'ensembles de motifs de formationTraitement des caractéristiques d’images ou de vidéos dans les espaces de caractéristiquesDispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant l’intégration et la réduction de données, p. ex. analyse en composantes principales [PCA] ou analyse en composantes indépendantes [ ICA] ou cartes auto-organisatrices [SOM]Séparation aveugle de source méthodes de Bootstrap, p. ex. "bagging” ou “boosting”
64.
MODEL GENERATION METHOD AND APPARATUS AND ELECTRONIC DEVICE
A model generation method and apparatus and an electronic device. The method comprises: acquiring a forward propagation computational graph of a target model, the forward propagation computational graph comprising a basic operator, and the basic operator being an operator having a primary mathematical operation (S101); performing differential transformation on the basic operator in the forward propagation computational graph to obtain a target computational graph, the target computational graph comprising at least one of a forward differential computational graph and a reverse differential computational graph (S102); and generating the target model on the basis of the target computational graph (S103).
Provided in the present disclosure are a neural network used for text recognition, a training method thereof and a text recognition method, relating to the field of artificial intelligence, and particularly relating to computer vision and deep learning technology. The neural network comprises: a first convolutional sub-network configured to output a first feature map on the basis of an image to be recognized; a local fusion sub-network configured to determine, on the basis of a feature vector of each pixel in the first feature map and a feature vector of a plurality of target pixels in the first feature map, local feature vectors of the pixels by means of a self-attention mechanism, so as to obtain a second feature map; a second convolutional sub-network configured to output a third feature map on the basis of the second feature map; a global fusion sub-network configured to determine, on the basis of a feature vector corresponding to each pixel in the third feature map and a feature vector of each pixel itself in the third feature map, global feature vectors of the pixels by using the self-attention mechanism, so as to obtain a fourth feature map; and an output sub-network configured to output a text recognition result on the basis of the fourth feature map.
G06V 30/18 - Extraction d’éléments ou de caractéristiques de l’image
G06V 30/19 - Reconnaissance utilisant des moyens électroniques
G06V 10/80 - Fusion, c.-à-d. combinaison des données de diverses sources au niveau du capteur, du prétraitement, de l’extraction des caractéristiques ou de la classification
G06V 10/82 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant les réseaux neuronaux
G06N 3/04 - Architecture, p. ex. topologie d'interconnexion
The present disclosure relates to the technical field of artificial intelligence, in particular to the technical field of natural language processing and deep learning, and provides a text processing method based on attention mechanism optimization, a network model training method and apparatus, a device, a medium, and a product. A specific implementation solution comprises: dividing M text statements in text to be processed to obtain N statement groups, N being an integer greater than 0, M being an integer not less than N, and each statement group in the N statement groups comprising at least one text statement; determining an attention distribution feature of at least one text statement in each statement group; and performing parallel operation on the basis of the attention distribution feature of each statement group to obtain an output result for the text to be processed.
The present disclosure relates to the field of artificial intelligence, and in particular to the field of deep learning and intelligent recommendation. Provided is a training method for a deep learning model. The specific implementation of the training method for a deep learning model comprises: according to first training data in the current training round, determining first target parameters, which need to be written into a target memory, from first network parameters required for performing embedding processing on the first training data, wherein the target memory is a memory contained in a target processor; according to first mapping relationships between storage slots of the target memory and network parameters, determining the remaining storage slots in the target memory; and in response to the remaining storage slots meeting a storage requirement of the first target parameters, writing the first target parameters into the target memory, such that a computing core contained in the target processor adjusts the first network parameters according to the first training data.
A batch task processing method and apparatus, and an electronic device. The method comprises: generating a batch task according to a task to be completed (101); executing the task to be completed in the batch task (102); and scheduling the task to be completed according to the completion progress of the task to be completed in the batch task (103).
Disclosed are a task search method and apparatus, a server and a storage medium. A specific implementation scheme comprises: acquiring a task scheduling strategy corresponding to an initial task, wherein the task scheduling strategy comprises a reasoning duration task scheduling strategy and a training task scheduling strategy, the reasoning duration task scheduling strategy is used for adjusting a reasoning duration corresponding to the initial task in a reasoning scenario, and the training task scheduling strategy is used for adjusting a search duration corresponding to the initial task in a training scenario; acquiring a training operation mode corresponding to the initial task; and searching the initial task by using the reasoning duration task scheduling strategy, the training task scheduling strategy and the training operation mode to obtain a target task.
A method for operating a deep learning compiler. The method comprises: acquiring a first operator set, which comprises a first operator and a second operator, of a deep learning framework, and acquiring a second operator set of a deep learning compiler; in response to the second operator set comprising the first operator, controlling the deep learning compiler to operate the first operator to obtain target data; transmitting the target data to the deep learning framework; and controlling the deep learning framework to operate the second operator in the first operator set on the basis of the target data.
The present disclosure relates to the technical field of computers, and particularly relates to the technical field of artificial intelligence such as deep learning. Provided is a method for processing an operator for a deep learning framework. The specific implementation solution comprises: acquiring an operator to be processed, wherein said operator comprises an operator kernel function, and template parameters unrelated to a deep learning framework; in response to input information for said operator being received, parsing the template parameters by using the input information, so as to obtain a plurality of complete template parameters related to the deep learning framework; and processing the operator kernel function according to the plurality of complete template parameters, so as to obtain an available operator for the deep learning framework. Further provided in the present disclosure are an apparatus for processing an operator for a deep learning framework, and an electronic device and a storage medium.
Provided in the present disclosure are a backbone network generation method and apparatus, a device and a storage medium, relating to the technical field of artificial intelligence, in particular to the technical field of deep learning and computer vision. A backbone network is applied to a visual processor, and the backbone network generation method comprises: acquiring calculation densities of a plurality of convolution kernels of different sizes; determining the convolution kernel having the maximum calculation density as a first convolution kernel; and generating a target backbone network on the basis of the first convolution kernel. In the backbone network generation method provided by the present disclosure, the backbone network can obtain a larger receptive field, relatively high reasoning speed and relatively high precision can be both achieved on the visual processor, and the cost is saved.
The present disclosure relates to the technical field of component management. Provided are an information generation method and apparatus, and a device and a storage medium. The method comprises: in response to it being determined that version information of a component has changed, determining the component to be a target component; on the basis of a component dependency relationship table, determining a dependency component, which depends on the target component, wherein the component dependency relationship table stores dependency relationships among components; on the basis of a preset version configuration rule, configuring version information for the dependency component; and on the basis of the relationship between the target component and the dependency component, storing, in the component dependency relationship table, changed version information of the target component and configured version information of the dependent component. The information generation method provided in the present disclosure automatically configures version information for a dependency component on the basis of a preset configuration rule, such that the efficiency of maintaining version information of a component is improved, thereby improving the efficiency of research and development.
A data processing method and apparatus, and a device and a storage medium. The data processing method comprises: acquiring a feature matrix, which is input by a target quantization network layer, and a weight matrix of the target quantization network layer (S101), wherein the number of columns of the feature matrix is equal to the number of rows of the weight matrix; according to a target segmentation coefficient of the target quantization network layer, dividing each row of feature elements of the feature matrix into at least two feature element sub-segments, and dividing each column of weight elements of the weight matrix into at least two weight element sub-segments (S102); and performing quantization processing on the at least two feature element sub-segments and the at least two weight element sub-segments according to the number of target quantization bits of the target quantization network layer, and determining an output feature of the target quantization network layer according to a quantization processing result (S103).
The present disclosure provides model inference methods and apparatus, devices and a storage medium, and relates to the technical field of data processing, in particular to the technical field of artificial intelligence. A method is applied to a GPU, and has a specific implementation solution as follows: receiving an operation kernel graph corresponding to a target neural network model sent by a CPU, nodes in the operation kernel graph respectively corresponding to operation kernels contained in the target neural network model, and the direction of each edge being used for representing a running sequence of operation kernels corresponding to nodes connected to said edge; after data to be processed sent by the CPU is received, according to the running sequence represented by the operation kernel graph, successively running the operation kernels to process said data so as to complete an inference process of the target neural network model; and feeding a model inference result back to the CPU. When the solution provided by the embodiments of the present disclosure is applied to model inference, the model inference efficiency of the GPU can be improved.
The present disclosure relates to the field of computers, and in particular to the field of big data, and provides a webpage processing method and apparatus for a website, an electronic device, and a storage medium. The specific implementation solution comprises: obtaining a plurality of webpages of a website, wherein the plurality of webpages are used for building the website; determining the weight of each webpage on the basis of the association relationship among the plurality of webpages, wherein the weight is used for representing the proportion of contribution that each webpage makes to the website; determining the website quality of the website on the basis of the weight of each webpage and the page quality of each webpage; and grading the website on the basis of the website quality, so as to obtain the quality grade of the website.
The present disclosure relates to the technical field of artificial intelligence, specifically relates to deep learning technology, and may be used in an image retrieval scenario. Provided are an image retrieval method and apparatus, and an electronic device, a storage medium and a computer program product. A specific implementation solution is as follows: obtaining, by means of a pre-trained global recall model, a global recall feature that takes both semantic information and visual information of an image to be retrieved into consideration; obtaining, by means of a pre-trained local check model, a local check feature, that is used for performing local feature point matching, of the image to be retrieved; and according to the global recall feature and the local check feature, determining, from a universal image library, an image similar to and/or the same as the image to be retrieved. The present disclosure provides universal retrieval logic for determining, from a universal image library, an image similar to or the same as an image to be retrieved, thereby improving the convenience and efficiency of image retrieval.
The present disclosure relates to fields of artificial intelligence such as deep learning and distributed storage, and provided thereby are a mixture-of-experts model implementation method and system, an electronic device, and a storage medium. The method may comprise: constructing a communication group, the communication group comprising a tensor parallel communication group, the tensor parallel communication group comprising at least two computing devices, and sparse parameters of computing devices in the same tensor parallel communication group using a tensor parallel segmentation means; and on the basis of the communication group, training a mixture-of-experts model. By applying the solution of the present disclosure, normal proceeding of model training can be guaranteed.
A model reasoning acceleration method and apparatus, and an electronic device and a storage medium, which relate to the field of artificial intelligence, and particularly relate to the field of deep learning. The specific implementation solution comprises: acquiring an original deep learning model which is obtained on the basis of dynamic language training (S101); acquiring a correspondence between a first description which is performed on a target object in the original deep learning model on the basis of a dynamic language and a second description which is performed on the target object in the original deep learning model on the basis of a predetermined static language (S102); on the basis of the correspondence, converting the target object in the original deep learning model into an object which is described on the basis of the predetermined static language, so as to obtain a target deep learning model (S103); and loading the target deep learning model into a predetermined deep learning reasoning optimizer, so as to obtain an optimized target deep learning model (S104).
The present disclosure provides a multi-core circuit, a data exchange method, an electronic device, and a storage medium. The multi-core circuit comprises: a public storage area; a plurality of interrupt registers; a plurality of status registers; and a plurality of computing cores. Each computing core is configured to: under the condition that a data exchange condition is satisfied, write first data into the public storage area, and trigger the interrupt registers corresponding to the computing cores other than the each computing core in the plurality of computing cores to generate interrupt signals; or under the condition that the interrupt register corresponding to each computing core generates an interrupt signal, read, according to a data read/write state of the public storage area recorded by the status register corresponding to each computing core, from the public storage area, second data written by the computing cores other than the each computing core in the plurality of computing cores.
Provided in the present disclosure are a speech recognition method and apparatus, and a device, a medium and a program product. The speech recognition method comprises: determining, by using an acoustic model, an acoustic score of at least one first candidate text unit corresponding to the current frame of audio data; according to a pre-established text pronunciation dictionary and the acoustic score of the at least one first candidate text unit, acquiring an acoustic score of at least one candidate pinyin corresponding to the current frame of audio data, wherein the text pronunciation dictionary is used for recording a correspondence between text and pinyin; according to a pre-established graph of pinyin and text, determining a language score of the at least one candidate pinyin by using a language model; and according to the acoustic score and the language score of the at least one candidate pinyin, determining a speech recognition result of the current frame of audio data.
G10L 15/06 - Création de gabarits de référenceEntraînement des systèmes de reconnaissance de la parole, p. ex. adaptation aux caractéristiques de la voix du locuteur
The present disclosure provides a voice wake-up model training method and apparatus, a voice wake-up method and apparatus, a device and a storage medium, and relates to the field of artificial intelligence, in particular to the fields of deep learning, voice technologies and the like. The specific implementation solution comprises: obtaining created voice recognition training data and voice wake-up training data, and first training a basic model according to the voice recognition training data to obtain model parameters of the basic model when a model loss function converges; then updating configuration parameters of a decoding module in the basic model on the basis of a model configuration instruction to obtain a first model; and training a first model according to the voice wake-up training data, and obtaining a trained voice wake-up model when the model loss function converges. According to the solution, the convergence speed of voice wake-up model training can be improved, audio data is processed and analyzed on the basis of the voice wake-up model, the recognition precision can be improved, and the false alarm rate is reduced.
G10L 17/04 - Entraînement, enrôlement ou construction de modèle
G10L 15/06 - Création de gabarits de référenceEntraînement des systèmes de reconnaissance de la parole, p. ex. adaptation aux caractéristiques de la voix du locuteur
G10L 15/16 - Classement ou recherche de la parole utilisant des réseaux neuronaux artificiels
G10L 15/22 - Procédures utilisées pendant le processus de reconnaissance de la parole, p. ex. dialogue homme-machine
G10L 17/02 - Opérations de prétraitement, p. ex. sélection de segmentReprésentation ou modélisation de motifs, p. ex. fondée sur l’analyse linéaire discriminante [LDA] ou les composantes principalesSélection ou extraction des caractéristiques
The present disclosure relates to the technical field of data processing, in particular, to big data and deep learning techniques in the technical field of artificial intelligence. Provided are a method and apparatus for updating the state of a point of interest (POI), and a device, a medium and a product. The specific implementation scheme involves: determining static features and dynamic features of a POI; on the basis of the static features and the dynamic features, predicting a first degree of health that represents the state of the POI; and updating the state of the POI on the basis of the latest state that matches the first degree of health. In the present disclosure, the state of a POI is updated by means of a degree of health that is predicted on the basis of static features and dynamic features of the POI, such that a capability of systematically updating the state of a POI is provided, and the quality of the POI is ensured in all aspects, thereby realizing the tracking of the state of the POI, and ensuring the application experience of a map.
Provided are a road condition prediction method and apparatus, and a corresponding model training method and apparatus, and a device and a medium, which relate to the technical fields of intelligent transportation, artificial intelligence, computer vision, etc. The implementation solution comprises: on the basis of a street view image of a designated road section at the current moment, mining visual features of the designated road section at the current moment (S101); on the basis of trajectory velocities of vehicles on the designated road section at the current moment and a plurality of successive historical moments prior to the current moment, mining trajectory velocity features of the designated road section at the current moment and the historical moments (S102); and on the basis of the visual features of the designated road section at the current moment and the trajectory velocity features of the designated road section at the current moment and the historical moments, predicting the road condition of the designated road section at the next moment (S103). The solution can effectively improve the accuracy of road condition prediction.
G06Q 10/04 - Prévision ou optimisation spécialement adaptées à des fins administratives ou de gestion, p. ex. programmation linéaire ou "problème d’optimisation des stocks"
G06Q 50/26 - Services gouvernementaux ou services publics
G06K 9/62 - Méthodes ou dispositions pour la reconnaissance utilisant des moyens électroniques
G06N 3/04 - Architecture, p. ex. topologie d'interconnexion
G06V 20/54 - Trafic, p. ex. de voitures sur la route, de trains ou de bateaux
G08G 1/01 - Détection du mouvement du trafic pour le comptage ou la commande
G08G 1/052 - Détection du mouvement du trafic pour le comptage ou la commande avec des dispositions pour déterminer la vitesse ou l'excès de vitesse
G08G 1/065 - Systèmes de commande du trafic pour véhicules routiers par comptage des véhicules dans une section de la route ou dans un parc de stationnement, c.-à-d. en comparant le nombre des arrivants et celui des partants
85.
INTELLIGENT INTERACTION METHOD AND APPARATUS, DEVICE, AND STORAGE MEDIUM
An intelligent interaction method and apparatus, a device, and a storage medium. The specific implementation solution comprises: S101: obtaining corresponding traffic information according to a detected interaction requirement; S102: generating interactive content according to the interaction requirement and a screening result of the traffic information; and S103: generating a driving parameter on the basis of the interactive content, the driving parameter comprising a parameter used for driving a digital human in an interactive interface to broadcast the interactive content.
Provided in the present disclosure are an image processing method and apparatus, a device and a storage medium, relating to the technical field of image processing, in particular to the technical field of computer vision, deep learning and natural language processing. The specific implementation solution is: acquiring a sample set to be processed, samples in the sample set comprising image content and label content; determining a first sample and at least one second sample for joint processing in the sample set; determining whether the first sample and the at least one second sample meet a joint processing exit condition; and in response to the fact that the first sample and the at least one second sample do not meet the joint processing exit condition, respectively carrying out image content joint and label content joint processing on the first sample and the at least one second sample, so as to obtain a target sample. The technical solution joins the first sample and the at least one second sample, so that the diversity of the sample set is improved, and the balance of the number of samples having different label content lengths is improved.
G06V 10/764 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant la classification, p. ex. des objets vidéo
87.
PRE-TRAINING METHOD AND MODEL FINE-TUNING METHOD FOR GEOGRAPHICAL PRE-TRAINING MODEL
The present application provides a pre-training method and model fine-tuning method for a geographical pre-training model, and relates to technical fields of artificial intelligence such as deep learning and graph structures. The pre-training method comprises: obtaining a sample node sequence, wherein the sample node sequence is generated on the basis of a preset point of interest heterogeneous graph and a random walk algorithm, the point of interest heterogeneous graph comprises nodes serving as points of interest and an edge connecting the nodes, names of the nodes are place names of corresponding points of interest, and the edge represents an association relationship that exists between corresponding nodes in the real world; inputting the sample node sequence which acts as a training sample into an initial geographical pre-training model; and controlling the initial geographical pre-training model to be trained according to a preset training target, and outputting a current geographical pre-training model that reaches the training target as a target geographical pre-training model. By means of integrating heterogeneous and multi-modal geographical knowledge into a model pre-training process, the effect of downstream tasks related to a geographical location is improved.
A method and apparatus for starting a model service, and a device, a medium and a product, which relate to the technical field of computers, and in particular, to the technical field of AI platforms. The method comprises: in response to a service model being triggered to start, acquiring a mirror file corresponding to the service model, wherein the mirror file comprises meta-information of the service model, and context information of a service process, which is run by the service model (S101); and loading the mirror file, so as to start the service model for a service (S102). By means of the method, a mirror file corresponding to a service model can be acquired, and the service model is started by loading the mirror file, such that the time for starting the model is shortened, and the speed for starting a model service is accelerated, thereby improving the user experience.
G06F 9/48 - Lancement de programmes Commutation de programmes, p. ex. par interruption
G06F 9/455 - ÉmulationInterprétationSimulation de logiciel, p. ex. virtualisation ou émulation des moteurs d’exécution d’applications ou de systèmes d’exploitation
89.
SCOREBOARD IMPLEMENTATION METHOD AND APPARATUS, SCOREBOARD, ELECTRONIC DEVICE, AND STORAGE MEDIUM
A scoreboard implementation method and apparatus, a scoreboard, an electronic device, and a storage medium, which relate to artificial intelligence fields such as artificial intelligence chips and intelligent speech. The method comprises: separately constructing M functional layers, where M is a positive integer greater than one, and different functional layers are respectively used for implementing different scoreboard functions (101); and using the M functional layers to form a scoreboard (102). The method is simple to implement and facilitates management and maintenance.
The present disclosure provides a method and apparatus for implementing mirror image storage of a memory model, and a storage medium, and relates to artificial intelligence chips, intelligent voice, and other artificial intelligence fields. The method may comprise: in response to a data update behavior occurring in a memory in a device under test, obtaining update behavior information of a current update, wherein the update behavior information is extracted from a bus; and according to the update behavior information, correspondingly updating data stored in a memory model. Effective memory verification and the like can be achieved by applying the solutions of the present disclosure.
Provided in the present disclosure are visual processing and model training methods, a device, a storage medium and a program product, relating to the technical field of computers, in particular to the field of computer vision and deep learning. A specific implementation solution is: by establishing an image classification network having the same backbone network as a visual model, carrying out self-supervised training on the image classification network by means of an unlabeled first data set; according to the weighting of the backbone network of the trained image classification network, initializing the weighting of the backbone network of the visual model, so as to obtain a pre-training model, the structure of the pre-training model being consistent with that of the visual model, and a real data set in a current computer vision task scenario being used for optimizing the weighting of the backbone network and more suitable for the current computer vision task; and training the pre-training model by means of a labeled second data set so as to obtain a trained visual model, thereby improving the accuracy of the visual model effectively.
G06V 10/764 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant la classification, p. ex. des objets vidéo
G06V 10/82 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant les réseaux neuronaux
G06K 9/62 - Méthodes ou dispositions pour la reconnaissance utilisant des moyens électroniques
The present disclosure relates to the field of artificial intelligence, e.g. intelligent speech and artificial intelligence chips, and provided are a speech chip implementation method, a speech chip, an intelligent speech product, an electronic device, and a storage medium. The method may comprise: constructing a speech chip comprising a first digital signal processor and a second digital signal processor, the first digital signal processor and the second digital signal processor corresponding to a same digital signal processor core identifier, and same utilizing a heterogeneous design means; and utilizing the speech chip to complete a chip processing function in a corresponding intelligent speech product, wherein the first digital signal processor and the second digital signal processor are separately utilized to complete different functions. By applying a solution of the present disclosure, implementation costs, etc. can be reduced.
The present disclosure relates to the technical field of artificial intelligence, and in particular to the technical fields of cloud platforms and machine learning. Provided are a method and apparatus for monitoring a model. The method comprises: acquiring data to be subjected to monitoring, wherein the data to be subjected to monitoring is at least part of running data of a target model to be monitored; performing classification processing on each piece of data in the data to be subjected to monitoring, so as to obtain a data feature respectively corresponding to each piece of the data, wherein different data features are features of data of a prediction result of the target model that is determined from different prediction dimensions, and one data feature corresponds to one prediction dimension; and monitoring the target model on the basis of data of data features corresponding to at least one prediction dimension, so as to obtain a monitoring result of the target model. By means of the method, the defect of lacking flexibility caused by relatively roughly monitoring a target model as a whole can be avoided, so as to improve the diversity and flexibility of monitoring, such that the monitoring of the target model is more targeted, thereby improving the effectiveness and reliability of monitoring.
A data prediction method and apparatus, an electronic device, a readable storage medium, and a computer program product, relating to the field of AI medical treatment. The specific solution comprises: determining a target image feature corresponding to an image under prediction, the image comprising an anterior segment of a target eye (S101); and using the target image feature and an intraocular lens feature to predict a postoperative vault after a target intraocular lens is implanted into the target eye, the intraocular lens feature being a feature preset for the target intraocular lens (S102).
The disclosure relates to the technical field of computers, in particular to the technical field of flow billing. Provided in the present disclosure are flow control and billing methods, apparatuses and system, an electronic device, a medium and a product, which may improve the accuracy of flow control in a high-concurrency scenario. A specific implementation solution is that: a flow control and billing module receives a begin request sent by a proxy service, and judges whether the sum of a current pre-billing amount, and a pre-billed amount and an expended amount corresponding to a user identifier is greater than an available credit so as to obtain a judgment result, wherein, the pre-billed amount corresponding to the user identifier is the sum of current pre-billing amounts of service requests which are being processed and which correspond to the user identifier; and then, whether to respond to a service request is determined on the basis of the judgment result, and the expended amounts corresponding to the user identifier are accumulated on the basis of the response result for the service request.
A character detection method and apparatus, an electronic device and a storage medium, relating to the field of artificial intelligence, and particularly relating to the field of deep learning, the field of character recognition and the field of image processing. A specific implementation solution of the character detection method comprises: detecting a binary image representing an area, where a character is located, in an image to be processed, so as to obtain contour information of said area (S210); on the basis of the contour information, determining position information of an initial bounding box for the character (S220); on the basis of the position information, determining an edge extension value for the initial bounding box (S230); and on the basis of the edge extension value, extending edges of the initial bounding box to obtain position information of a bounding box for recognizing the character (S240).
G06K 9/00 - Méthodes ou dispositions pour la lecture ou la reconnaissance de caractères imprimés ou écrits ou pour la reconnaissance de formes, p.ex. d'empreintes digitales
97.
CONTENT INITIALIZATION METHOD AND APPARATUS, ELECTRONIC DEVICE AND STORAGE MEDIUM
The present disclosure relates to the technical field of computers, and particularly to the field of deep learning and the field of distributed computing. Provided in the present disclosure are a content initialization method and apparatus, an electronic device and a storage medium. The content initialization method is applied to any device amongst a plurality of devices included in a distributed system. A specific implementation solution of the content initialization method is: according to the size information of a resource space for a distributed system and the identification information of any device, determining the spatial information of a first subspace in the resource space for the device, the spatial information comprising the position information of the first subspace with respect to the resource space; and according to a random seed and the position information, determining initialization content for the first subspace.
The present disclosure provides information search methods and apparatuses, an electronic device, a storage medium and a computer program product, and relates to the technical field of artificial intelligence such as deep learning, big data, intelligent search, etc. A specific solution comprises: a server acquiring a search keyword of a first client; acquiring a vector representation group of at least one candidate target, wherein the vector representation group comprises vector representations of the first client and at least one second client for the same candidate target; according to the search keyword and the vector representation group of the at least one candidate target, singling out at least one search target matched with the search keyword from the at least one candidate target; and sending the at least one search target to the first client, so that the first client acquires a search result from a database of the first client according to the at least one search target. Without involving any private data, the technical solution of the present disclosure may perform a search by using the data of other nodes, so that a search effect is enhanced.
A method and apparatus for adjusting playback volume, and a device, a storage medium, and a program product, which relate to the technical field of artificial intelligence, and in particular to the technical field of voice. The method comprises: in response to receiving an original voice, determining a target voice segment in the original voice, the target voice segment comprising a keyword (S210); determining, by using a deep learning model, whether the target voice segment comprises breathy voice (S220); and in the case that the target voice segment comprises the breathy voice, adjusting the playback volume to be below predetermined volume (S230), wherein the deep learning model is trained according to the following method: determining a breathy voice likelihood corresponding to a voice sample by using the deep learning model (S310); determining a loss value according to the breathy voice likelihood and a standard likelihood corresponding to the voice sample (S320); and adjusting parameters of the deep learning model according to the loss value (S330).
An audio data processing method and apparatus, an electronic device, a computer readable storage medium, and a computer program product, relating to the field of artificial intelligence, and in particular to an audio processing technology. The method comprises: acquiring human voice audio data to be adjusted (201) and acquiring reference human voice audio data (202), wherein the reference human voice audio data and the human voice audio data to be adjusted are obtained on the basis of the same text information; respectively framing the human voice audio data to be adjusted and the reference human voice audio data to respectively obtain a first audio frame set and a second audio frame set (203); respectively recognizing pronunciation units corresponding to audio frames (204); on the basis of timestamps of the audio frames, respectively determining timestamps of pronunciation units in the human voice audio data to be adjusted and the reference human voice audio data (205); and adjusting the timestamps of the pronunciation units in the human voice audio data to be adjusted, so that the timestamps of the corresponding pronunciation units in the human voice audio data to be adjusted and the reference human voice audio data are consistent (206).
G10L 25/03 - Techniques d'analyse de la parole ou de la voix qui ne se limitent pas à un seul des groupes caractérisées par le type de paramètres extraits
G10L 25/51 - Techniques d'analyse de la parole ou de la voix qui ne se limitent pas à un seul des groupes spécialement adaptées pour un usage particulier pour comparaison ou différentiation
brevets
