WO2024255723A1 - 用于多模态数据的文本分类方法及装置、制冷设备、介质 - Google Patents
用于多模态数据的文本分类方法及装置、制冷设备、介质 Download PDFInfo
- Publication number
- WO2024255723A1 WO2024255723A1 PCT/CN2024/098340 CN2024098340W WO2024255723A1 WO 2024255723 A1 WO2024255723 A1 WO 2024255723A1 CN 2024098340 W CN2024098340 W CN 2024098340W WO 2024255723 A1 WO2024255723 A1 WO 2024255723A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- data
- text
- image
- speech
- information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L13/00—Speech synthesis; Text to speech systems
- G10L13/08—Text analysis or generation of parameters for speech synthesis out of text, e.g. grapheme to phoneme translation, prosody generation or stress or intonation determination
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
- G06N3/0455—Auto-encoder networks; Encoder-decoder networks
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/096—Transfer learning
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Definitions
- the present application relates to the field of data processing technology, for example, to a text classification method and device, refrigeration equipment, and medium for multimodal data.
- the related technology discloses a text classification method based on multimodal deep learning, including: S01, obtaining real-time audio and video data and historical audio and video data. S02, preprocessing the real-time audio and video data and historical audio and video data to obtain valid voice data and video data. S03, transcribing the valid voice data into voice text data. S04, obtaining the video image of the local area in the valid video data, and transcribing the video image into image text data. S05, according to the voice text data and the image text data, obtaining the context information of the text data and the weight information of the text semantic features.
- S06 after combining the context information and the weight information through the fully connected layer, outputting them to the classifier to calculate the score to obtain the classification result information, and judging the type information of the text generated by the audio and video data.
- S07 outputting the category information of the generated text.
- the present disclosure provides a method, device, refrigeration equipment and medium for text classification of multimodal data. To solve the problem of improving the effectiveness of short text extraction from multimodal data.
- the method includes: obtaining text information, voice information and image information based on multimodal data; obtaining a voice text feature vector and an image text feature vector based on the voice information and image information; performing text preprocessing on the text information, the voice text feature vector and the image text feature vector to obtain short text data.
- obtaining a speech-text feature vector and an image-text feature vector based on speech information and image information includes: preprocessing the speech information to obtain valid speech data and valid video data; performing speech recognition processing on the valid speech data to obtain a speech-text feature vector; and obtaining an image-text feature vector based on the valid video data and image information.
- obtaining an image text feature vector based on valid video data and image information includes: performing frame extraction processing on the valid video data to obtain second image data; performing filtering processing on the image information to obtain first image data; performing image synthesis processing on the first image data and the second image data to obtain image data; and inputting the image data into a first deep neural network model for model training to obtain an image text feature vector.
- the first deep neural network model includes a migration and distillation diffusion deep fusion model.
- the speech recognition processing of the valid speech data to obtain the speech text feature vector includes: inputting the valid speech data to the encoder for speech encoding to obtain the speech encoding data; inputting the speech encoding data to the decoder for speech decoding to obtain the speech text feature vector.
- the text information, speech text feature vectors and image text feature vectors are subjected to text preprocessing to obtain short text data, including: extracting short text information of each of the text information, speech text feature vectors and image text feature vectors; and aggregating short text information of each of the text information, speech text feature vectors and image text feature vectors to obtain short text data.
- the method further includes: performing data annotation processing on the short text data to obtain the data to be inspected; when the data to be inspected meets the quality inspection requirements, using the trained Bert model to extract text features of the data to be inspected to obtain target short text feature data; using the Focal Loss loss function to perform short text classification processing on the target short text feature data to obtain the target short text data.
- the Bert model is trained in the following manner: the initial Bert model is trained with the benchmark network parameters to obtain the network parameters and sorting information after the model training; the target number of network parameters is determined according to the network parameters and sorting information after the model training; the initial Bert model is trained with the target number of network parameters to obtain a trained Bert model.
- the method of training the initial Bert model using the target number of network parameters to obtain a trained Bert model includes: training the initial Bert model using the target number of network parameters to obtain a baseline Bert model; and migrating the teacher network model to the student network model using knowledge distillation technology to obtain a baseline student network model.
- the teacher network model is transferred to the student network model using the knowledge distillation technology to obtain After the benchmark student network model, it also includes: obtaining the number of layers of all encoders of the benchmark student network model and the weight of each encoder; determining the target number of layers of the encoder according to the number of layers of all encoders and the weight of each encoder; using the encoder of the target number of layers to train the benchmark student network model to obtain a trained Bert model.
- the device includes: a speech preprocessing unit, configured to preprocess speech information associated with multimodal data to obtain valid speech data; a model training unit, configured to perform model training on image data using a first deep neural network model to obtain an image text feature vector; a speech recognition unit, configured to perform speech recognition processing on valid speech data to obtain a speech text feature vector; and a text preprocessing unit, configured to perform text preprocessing on text information, speech text feature vectors, and image text feature vectors to obtain short text data.
- a speech preprocessing unit configured to preprocess speech information associated with multimodal data to obtain valid speech data
- a model training unit configured to perform model training on image data using a first deep neural network model to obtain an image text feature vector
- a speech recognition unit configured to perform speech recognition processing on valid speech data to obtain a speech text feature vector
- a text preprocessing unit configured to perform text preprocessing on text information, speech text feature vectors, and image text feature vectors to obtain short text data.
- the voice preprocessing unit is also configured to preprocess the audio and video information to obtain valid video data
- the device also includes: an image frame extraction unit, configured to perform frame extraction processing on the valid video data to obtain second image data; an image filtering unit, configured to filter processing on the image information to obtain first image data; and an image fusion unit, configured to perform image synthesis processing on the first image data and the second image data to obtain image data.
- the device also includes: a data labeling unit, configured to perform data labeling processing on the short text data to obtain the data to be inspected; a data quality inspection unit, configured to use the trained Bert model to perform text feature extraction on the data to be inspected when the data to be inspected meets the quality inspection requirements to obtain target short text feature data; a text classification unit, configured to use the Focal Loss loss function to perform short text classification processing on the target short text feature data to obtain the target short text data.
- a data labeling unit configured to perform data labeling processing on the short text data to obtain the data to be inspected
- a data quality inspection unit configured to use the trained Bert model to perform text feature extraction on the data to be inspected when the data to be inspected meets the quality inspection requirements to obtain target short text feature data
- a text classification unit configured to use the Focal Loss loss function to perform short text classification processing on the target short text feature data to obtain the target short text data.
- the device includes: a processor and a memory storing program instructions, and the processor is configured to execute the aforementioned text classification method for multimodal data when running the program instructions.
- the refrigeration equipment includes: an equipment body; and the aforementioned text classification device for multimodal data is installed on the equipment body.
- the storage medium stores program instructions, and when the program instructions are run, they execute the aforementioned text classification method for multimodal data.
- the text classification method, device, refrigeration equipment and storage medium for multimodal data provided by the embodiments of the present disclosure can achieve the following technical effects:
- the disclosed embodiment obtains text information, voice information, and image information based on multimodal data. Then, based on the voice information and image information, a voice-text feature vector and an image-text feature vector are obtained, so as to respectively reflect the text features of the voice and image using the voice-text feature vector and the image-text feature vector. Finally, the text information, the voice-text feature vector, and the image-text feature vector are preprocessed to obtain short text data. In this way, by preprocessing the voice-text feature vector and the image-text feature vector that carry text features with the text information, the disclosed embodiment can extract short text features from different data sources from the multimodal data, thereby obtaining high-quality short texts, thereby improving the quality of the short text. Effectiveness of short text extraction from multimodal data.
- FIG1 is a schematic diagram of a text classification device for multimodal data provided by an embodiment of the present disclosure
- FIG2 is a schematic diagram of a text classification method for multimodal data provided by an embodiment of the present disclosure
- FIG3 is a schematic diagram of another text classification method for multimodal data provided by an embodiment of the present disclosure.
- FIG4 is a schematic diagram of another text classification method for multimodal data provided by an embodiment of the present disclosure.
- FIG5 is a schematic diagram of another text classification method for multimodal data provided by an embodiment of the present disclosure.
- FIG6 is a schematic diagram of another text classification method for multimodal data provided by an embodiment of the present disclosure.
- FIG7 is a schematic diagram of an application of an embodiment of the present disclosure.
- FIG8 is a schematic diagram of a text classification device for multimodal data provided by an embodiment of the present disclosure.
- FIG9 is a schematic diagram of another text classification device for multimodal data provided by an embodiment of the present disclosure.
- FIG. 10 is a schematic diagram of a refrigeration device provided in an embodiment of the present disclosure.
- 10 speech preprocessing unit
- 20 image filtering unit
- 30 speech recognition unit
- 40 image frame extraction unit
- 50 Image fusion unit
- 60 Model training unit
- 70 Text preprocessing unit
- 80 Data annotation unit
- 90 data quality inspection unit
- 100 text classification unit
- 200 a text classification device for multimodal data
- 300 a text classification device for multimodal data
- 400 processor; 401: memory; 402: communication interface; 403: bus; 600: refrigeration equipment.
- the character "/" indicates that the preceding and following objects are in an "or" relationship.
- A/B indicates: A or B.
- a and/or B means: A or B, or, A and B.
- correspondence may refer to an association relationship or a binding relationship.
- correspondence between A and B means that there is an association relationship or a binding relationship between A and B.
- an embodiment of the present disclosure provides a text classification device 200 for multimodal data, including a speech preprocessing unit 10, a model training unit 60, a speech recognition unit 30, and a text preprocessing unit 70.
- the speech preprocessing unit 10 is configured to preprocess the speech information associated with the multimodal data to obtain valid speech data.
- the model training unit 60 is configured to perform model training on the image data using a first deep neural network model to obtain an image text feature vector.
- the speech recognition unit 30 is configured to perform speech recognition processing on the valid speech data to obtain a speech text feature vector.
- the text preprocessing unit 70 is configured to perform text preprocessing on the text information, the speech text feature vector, and the image text feature vector to obtain short text data.
- the speech preprocessing unit 10 is also configured to preprocess the audio and video information to obtain valid video data.
- the text classification system for multimodal data also includes an image frame extraction unit 40, an image filtering unit 20, and an image fusion unit 50.
- the image frame extraction unit 40 is configured to perform frame extraction processing on the valid video data to obtain second image data.
- the image filtering unit 20 is configured to filter the image information to obtain first image data.
- the image fusion unit 50 is configured to perform image synthesis processing on the first image data and the second image data to obtain image data.
- the text classification system for multimodal data further includes a data annotation unit 80, a data quality inspection unit 90, and a text classification unit 100.
- the data annotation unit 80 is configured to perform data annotation processing on the short text data to obtain the data to be inspected.
- the data quality inspection unit 90 is configured to use the trained Bert model to extract text features from the data to be inspected when the data to be inspected meets the quality inspection requirements to obtain target short text feature data.
- the text classification unit 100 is configured to use the Focal Loss loss function to perform short text classification processing on the target short text feature data to obtain the target short text data.
- an embodiment of the present disclosure provides a text classification method for multimodal data, including:
- the processor obtains text information, voice information and image information according to multimodal data.
- the types of multimodal data include part or all of text information, voice information, and image information.
- the processor obtains the multimodal data in the following manner: The processor obtains the multimodal data through a human-computer interaction mode.
- the human-computer interaction mode includes but is not limited to Bluetooth, Wi-Fi (Wireless Fidelity), Internet, Web (World Wide Web), microphone array, pickup, and software.
- the software includes One or more of APP (Application), information collection tool and touch screen.
- the processor obtains a speech text feature vector and an image text feature vector according to the speech information and the image information.
- the processor performs text preprocessing on the text information, the speech text feature vector and the image text feature vector to obtain short text data.
- the embodiment of the present disclosure obtains text information, voice information and image information based on the multimodal data. Then, based on the voice information and image information, a voice text feature vector and an image text feature vector are obtained, so as to respectively reflect the text features of the voice and image using the voice text feature vector and the image text feature vector. Finally, the text information, the voice text feature vector and the image text feature vector are preprocessed to obtain short text data.
- the processor extracts text information from the multimodal data in the following manner: the processor obtains initial text data from the multimodal data.
- the processor performs text preprocessing on the initial text data to obtain text information.
- the text preprocessing includes word segmentation, word form normalization, and deletion of stop words.
- the processor obtains a speech text feature vector and an image text feature vector according to the speech information and the image information, including:
- the processor pre-processes the voice information to obtain valid voice data and valid video data.
- the processor pre-processes the voice information to obtain valid voice data and valid video data, including:
- the processor performs data cleaning and format conversion processing on the multimodal data in turn to extract valid sound data.
- the processor performs voice and video separation processing on the valid sound data to obtain voice preprocessing data and valid video data.
- the processor performs effective duration processing on the voice preprocessing data to obtain valid voice data. In this way, by performing data cleaning and format conversion processing, some interference in the multimodal data can be removed and the data format can be unified to obtain valid sound data.
- voice and video separation processing on the valid sound data By performing voice and video separation processing on the valid sound data, the voice preprocessing data and the valid video data can be separated to obtain the voice preprocessing data.
- By performing effective duration processing on the voice preprocessing data empty voice and background sound can be effectively processed or eliminated. Among them, the background sound includes the machine sound during the machine operation stage.
- the processor can perform data cleaning, format conversion processing and voice and video separation processing on the multimodal data through data processing tools or data processing scripts.
- the processor performs speech recognition processing on the valid speech data to obtain a speech text feature vector.
- the processor performs speech recognition processing on the valid speech data to obtain a speech text feature vector, including: the processor inputs the valid speech data to the first encoder for encoding processing to obtain speech encoding data.
- the processor uses the first
- the second deep neural network model extracts features from the speech coding data to obtain a speech text feature vector.
- the first encoder is a Wav2VEC encoder.
- the second deep neural network is configured with a multi-head attention mechanism. In this way, by adopting the second deep neural network configured with a multi-head attention mechanism, the vector space mapping of the speech coding data is realized, and the semantic, grammatical and contextual features covered by the speech are extracted and reflected through the speech text feature vector.
- the processor obtains an image text feature vector according to the valid video data and image information.
- the embodiment of the present disclosure can extract valid voice data and valid video data from the voice information by preprocessing the voice information, reduce interference in multimodal data and achieve the unification of data format. In addition, empty voice and background sound can be effectively eliminated. The effectiveness of the recognition of valid voice data and valid video data is improved.
- the embodiment of the present disclosure performs voice recognition processing on the valid voice data to obtain a voice-text feature vector. This is conducive to fully mining the voice-text features from the voice. And based on the valid video data and image information, an image-text feature vector is obtained, which is conducive to fully mining the image-text features from the image.
- the processor obtains an image text feature vector according to valid video data and image information, including:
- the processor performs frame extraction processing on the effective video data to obtain second image data.
- S22 The processor filters the image information to obtain first image data.
- the processor performs image synthesis processing on the first image data and the second image data to obtain image data.
- the processor inputs the image data into the first deep neural network model for model training to obtain the image text feature vector.
- the embodiment of the present disclosure first extracts the second image data from the video data by performing frame extraction on the effective video data, and performs filtering processing on the image information to extract the first image data from the image information. Then, the first image data and the second image data are subjected to image synthesis processing to effectively fuse the first image data and the second image to generate image data. Finally, the fused image data is input into the first deep neural network model for model training, which is conducive to accurately identifying semantic features from the image data.
- the first deep neural network model includes a migration and distillation diffusion deep fusion model.
- the migration and distillation diffusion deep fusion model means using migration learning to use the output of a pre-trained complex model as a supervisory signal to train a simple network model.
- the complex model is called the teacher model.
- the simple network model is called the student model.
- the temporal, spatial and long-distance dependency characteristics of the model can be utilized to introduce the teacher model into the temporally and spatially continuous student model, which is more conducive to mining rich semantic features from image data and further improving the effectiveness of image text feature vectors.
- the processor performs speech recognition processing on the valid speech data to obtain a speech text feature vector, including:
- the embodiment of the present disclosure performs speech encoding by inputting valid speech data into the encoder, so that the speech encoding data obtained by encoding does not lose useful speech data, which is beneficial to the subsequent extraction of speech text features.
- the processor inputs valid voice data to the encoder for voice encoding to obtain voice encoding data, including: the processor uses a Wav2VEC encoder to encode the valid voice data to obtain the voice encoding data.
- the processor uses a Wav2VEC encoder to encode the valid voice data to obtain the voice encoding data.
- the elimination effect of redundant voice data can be enhanced, thereby obtaining purer voice encoding data.
- the processor inputs the speech coding data to the decoder for speech decoding to obtain the speech text feature vector, including: the processor inputs the speech coding signal to the Transformer network model for decoding processing to obtain the speech text feature vector.
- the processor inputs the speech coding signal to the Transformer network model for decoding processing to obtain the speech text feature vector.
- the processor performs text preprocessing on the text information, the speech text feature vector, and the image text feature vector to obtain short text data, including:
- the processor extracts short text information of each of the text information, the speech text feature vector, and the image text feature vector.
- the processor collects the short text information of the text information, the speech text feature vector and the image text feature vector to obtain the short text data.
- the disclosed embodiment extracts short text information from text information, speech text feature vectors and image text feature vectors respectively, and then aggregates the short text information from the above different data sources. Effective extraction of short text features from different data sources is achieved, which is conducive to obtaining high-quality short texts, thereby improving the effectiveness of short text extraction from multimodal data.
- the present disclosure also provides a text classification method for multimodal data, including:
- the processor obtains text information, voice information and image information according to the multimodal data.
- the processor obtains a speech text feature vector and an image text feature vector according to the speech information and the image information.
- the processor performs text preprocessing on the text information, the speech text feature vector and the image text feature vector to obtain short text data.
- the processor performs data annotation processing on the short text data to obtain data to be inspected.
- the processor performs data labeling on the short text data to obtain the data to be inspected, including: the processor labels the slots and metadata attribute labels on the short text data to obtain the data to be inspected, thus providing a basis for subsequent data quality inspection.
- the processor uses the trained Bert model to extract text features from the data to be inspected to obtain target short text feature data.
- the processor determines that the data to be processed meets the quality inspection requirements in the following manner: the metadata attribute label and the food attribute label are correct and the error rate is less than or equal to 2%.
- the processor uses the Focal Loss loss function to perform short text classification processing on the target short text features to obtain the target short text data.
- the embodiment of the present disclosure sequentially performs data annotation and data quality inspection, and when the quality inspection requirements are met, uses the trained Bert model to extract text features from the data to be inspected, and obtains the target short text feature data, which is conducive to extracting short text feature data containing rich short text features from the data to be inspected. Since there is a problem of type imbalance in short text feature data, the classification accuracy of short text classification processing is affected.
- the embodiment of the present disclosure uses the Focal Loss loss function to perform short text classification processing on the target short text features, so as to reduce the impact of the type imbalance of short text feature data on the accuracy of short text classification processing, and further ensure the quality of short texts.
- the processor uses the Focal Loss loss function to perform short text classification processing on the target short text feature to obtain target short text data, including: the processor uses the Focal Loss loss function to perform short text classification processing on the target short text feature data to obtain short text classification data.
- the processor performs data enhancement processing on the short text classification data to obtain target short text data. In this way, by performing data enhancement processing on the short text classification data, the limited short text classification data set can be expanded, the training effect of the Bert model can be improved, and the generalization ability of the model can be enhanced.
- the processor trains the Bert model in the following manner:
- the processor performs model training on the initial Bert model with the reference network parameters to obtain the network parameters and sorting information after the model training.
- the processor determines the target number of network parameters according to the network parameters after model training and the sorting information.
- the processor determines the target number of network parameters according to the network parameters after model training and the sorting information, including: the processor selects the number of network parameters before sorting ⁇ as the target number according to the sorting information.
- the processor trains the initial Bert model using the target number of network parameters to obtain a trained Bert model.
- the number of network parameters of the initial Bert model is intercepted to achieve fine-tuning of the initial Bert model, retaining the number of network layers and network parameters with high model weights, which can reduce the memory space occupied by the model.
- the processor obtains the reference network parameters in the following manner:
- the processor extracts training data from the target short text data.
- the processor uses the training data to perform model training on the initial Bert model to obtain benchmark network parameters.
- the processor trains an initial BERT model using a target number of network parameters to obtain a trained BERT model, including:
- the processor trains the initial Bert model with the target number of network parameters to obtain a benchmark Bert model.
- the processor uses knowledge distillation technology to migrate the teacher network model to the student network model to obtain a benchmark student network model.
- the benchmark Bert model is the teacher network model.
- the benchmark student network model is a trained Bert model. In this way, by migrating the teacher network model to the student network model, the large model is used to guide the small model for model training.
- the obtained benchmark student network model has performance comparable to that of the teacher network model.
- the number of network parameters is greatly reduced to achieve model compression and acceleration.
- the processor uses knowledge distillation technology to migrate the teacher network model to the student network model, and obtains the benchmark student network model, it also includes: the processor obtains the number of layers of all encoders of the benchmark student network model and the weight of each encoder.
- the processor determines the target number of layers of the encoder based on the number of layers of all encoders and the weight of each encoder.
- the processor trains the benchmark student network model using the encoder of the target number of layers to obtain a trained Bert model.
- the number of encoder layers of the pruned Bert model is reduced, which can reduce the model's occupancy of memory space and improve the efficiency of model training of the Bert model.
- the processor determines the target number of layers of the encoder based on the number of layers of all encoders and the weight of each encoder, including: the processor arranges the weights of all encoders in descending order according to the weight size.
- the processor selects the number of layers of the encoders in the first ⁇ positions from the weights arranged in descending order as the target number of layers.
- the text classification method for multimodal data includes the following steps:
- S101 The processor collects multimodal data.
- S102 The processor performs data cleaning on the multimodal data to obtain voice information, text information and image information.
- S103 The processor pre-processes the voice information to obtain valid voice data and valid video data.
- S1041 The processor inputs valid voice data and passes it through the Wav2 encoder and the Transformer decoder in sequence to obtain a voice text feature vector.
- S1042 The processor extracts image frames from the valid video data to obtain a second target image.
- S1043 The processor performs image filtering on the image information to obtain a first target image.
- S1044 The processor performs image synthesis on the second target image and the first target image to obtain image data.
- the processor inputs the image data to the migration and distillation diffusion deep fusion model to perform vector space mapping processing to extract image text feature vectors from the image data.
- S105 The processor performs text preprocessing on the text information, the speech text feature vector and the image text feature vector to obtain short text data.
- S106 The processor performs data annotation processing on the short text data to obtain data to be inspected.
- S107 The processor determines whether the data to be inspected meets the quality inspection requirements. If so, execute S108; otherwise, execute S109.
- S108 The processor outputs the data to be inspected as a text classification result.
- S109 The processor uses the trained Bert model to extract text features from the quality inspection data to obtain target short text feature data.
- the processor uses the Focal Loss function to perform short text classification on the target short text features, obtains the target short text data, and outputs the target short text data as the text classification result.
- the embodiment of the present disclosure provides a text classification device 300 for multimodal data, including a processor 400 and a memory 401.
- the device may also include a communication interface 402 and a bus 403.
- the processor 400, the communication interface 402, and the memory 401 may communicate with each other through the bus 403.
- the communication interface 402 may be used for information transmission.
- the processor 400 may call the logic instructions in the memory 401 to execute the text classification method for multimodal data of the above embodiment.
- logic instructions in the memory 401 described above may be implemented in the form of software functional units and when sold or used as independent products, may be stored in a computer-readable storage medium.
- the memory 401 is a computer-readable storage medium that can be used to store software programs and computer executable programs, such as program instructions/modules corresponding to the method in the embodiment of the present disclosure.
- the processor 400 executes the function application and data processing by running the program instructions/modules stored in the memory 401, that is, the text classification method for multimodal data in the above embodiment is implemented.
- the memory 401 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application required for at least one function; the data storage area may store data created according to the use of the terminal device, etc.
- the memory 401 may include a high-speed random access memory and may also include a non-volatile memory.
- an embodiment of the present disclosure provides a refrigeration device 600, including: a device body, and the above-mentioned text classification device 200 (300) for multimodal data.
- the text classification device 200 (300) for multimodal data is installed on the device body.
- the installation relationship described here is not limited to placement inside the product, but also includes installation connections with other components of the product, including but not limited to physical connections, electrical connections, or signal transmission connections. It can be understood by those skilled in the art that the text classification device 200 (300) for multimodal data can be adapted to a feasible product body, thereby realizing other feasible embodiments.
- An embodiment of the present disclosure provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured to execute the above-mentioned text classification method for multimodal data.
- the embodiments of the present disclosure provide a computer program.
- the computer program When the computer program is executed by a computer, the computer implements the above-mentioned text classification method for multimodal data.
- the present disclosure provides a computer program product, wherein the computer program product includes a computer program product stored in a computer program Reading computer instructions on a storage medium, when the program instructions are executed by a computer, enables the computer to implement the above-mentioned text classification method for multimodal data.
- the computer-readable storage medium mentioned above may be a transient computer-readable storage medium or a non-transitory computer-readable storage medium.
- the technical solution of the embodiment of the present disclosure can be embodied in the form of a software product, which is stored in a storage medium and includes one or more instructions for enabling a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the embodiment of the present disclosure.
- the aforementioned storage medium may be a non-transient storage medium, including: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes, or a transient storage medium.
- the term “and/or” as used in this application refers to any and all possible combinations of listings containing one or more associated ones.
- the term “comprise” and its variants “comprises” and/or comprising refer to the presence of stated features, wholes, steps, operations, elements, and/or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and/or groups thereof.
- the elements defined by the sentence “comprising a " do not exclude the presence of other identical elements in the process, method or device comprising the elements.
- each embodiment may focus on the differences from other embodiments, and the same and similar parts between the various embodiments may refer to each other.
- the relevant parts can refer to the description of the method part.
- the disclosed methods and products can be implemented in other ways.
- the device embodiments described above are only schematic.
- the division of the units can be only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.
- the coupling or direct coupling or communication connection between each other shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
- each functional unit in the embodiment of the present disclosure may be integrated in a processing unit, or each unit may exist physically alone, or two or more units may be integrated in one unit.
- each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function.
- the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Databases & Information Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
本申请涉及数据处理技术领域,公开一种用于多模态数据的文本分类方法,包括:根据多模态数据,获得文本信息、语音信息以及图像信息;根据语音信息以及图像信息,获得语音文本特征向量以及图像文本特征向量;对文本信息、语音文本特征向量以及图像文本特征向量进行文本预处理,获得短文本数据。本申请还公开一种用于多模态数据的文本分类装置及制冷设备、介质。
Description
本申请基于申请号为202310694668.X、申请日为2023年6月12日的中国专利申请提出,并要求该中国专利申请的优先权,该中国专利申请的全部内容在此引入本申请作为参考。
本申请涉及数据处理技术领域,例如涉及一种用于多模态数据的文本分类方法及装置、制冷设备、介质。
目前,随着多模态语音降噪技术、语音识别技术和信号处理技术的快速发展,从多模态数据中提取出用以实现交互的语音或者文本成为工业界的研究热点。在智能家居场景下,针对用户交互场景,智能认知与感知存在交互慢和反馈信息错误率高的问题,影响用户的产品体验。因此,如何从多模态数据中提取获得高质量的短文本,成为当前亟需解决的技术难题。
相关技术公开一种基于多模态深度学习的文本分类方法,包括:S01,获取实时音视频数据和历史音视频数据。S02,对实时音视频数据和历史音视频数据进行预处理,获取有效的语音数据和视频数据。S03,转写有效语音数据为语音文本数据。S04,获取有效视频数据中局部区域的视频图像,并转写视频图像为图像文本数据。S05,根据语音文本数据和图像文本数据,获得该文本数据的上下文信息和文本语义特征的权重信息。S06,将上下文信息和权重信息经全连接层组合后,输出至分类器计算得分得到分类结果信息,并判断音视频数据生成文本的类型信息。S07,输出生成文本的类别信息。
在实现本公开实施例的过程中,发现相关技术中至少存在如下问题:
相关技术选取实时音视频数据和历史音视频数据作为参考数据,通过参考数据生成交互场景的文本。虽然实时音视频数据和历史音视频数据包含有文本特征,但是,多模态数据的数据来源不限于音视频数据。如此,仅采用音视频数据作为文本提取的参考数据,无法充分挖掘出多模态数据中其他数据来源的短文本特征,影响短文本提取的有效性。
需要说明的是,在上述背景技术部分公开的信息仅用于加强对本申请的背景的理解,因此可以包括不构成对本领域普通技术人员已知的现有技术的信息。
发明内容
为了对披露的实施例的一些方面有基本的理解,下面给出了简单的概括。所述概括不是泛泛评述,也不是要确定关键/重要组成元素或描绘这些实施例的保护范围,而是作为后面的详细说明的序言。
本公开实施例提供了一种用于多模态数据的文本分类方法、装置、制冷设备和介质,
以解决提升多模态数据短文本提取的有效性。
在一些实施例中,所述方法包括:根据多模态数据,获得文本信息、语音信息以及图像信息;根据语音信息以及图像信息,获得语音文本特征向量以及图像文本特征向量;对文本信息、语音文本特征向量以及图像文本特征向量进行文本预处理,获得短文本数据。
在一些实施例中,所述根据语音信息以及图像信息,获得语音文本特征向量以及图像文本特征向量,包括:对语音信息进行预处理,获得有效语音数据以及有效视频数据;对有效语音数据进行语音识别处理,获得语音文本特征向量;根据有效视频数据以及图像信息,获得图像文本特征向量。
在一些实施例中,所述根据有效视频数据以及图像信息,获得图像文本特征向量,包括:对有效视频数据进行帧提取处理,获得第二图像数据;对图像信息进行过滤处理,获得第一图像数据;对第一图像数据以及第二图像数据进行图像合成处理,获得图像数据;输入图像数据至第一深度神经网络模型进行模型训练,获得图像文本特征向量。
在一些实施例中,所述第一深度神经网络模型包括迁移与蒸馏扩散深度融合模型。
在一些实施例中,所述对有效语音数据进行语音识别处理,获得语音文本特征向量,包括:输入有效语音数据至编码器进行语音编码,获得语音编码数据;输入语音编码数据至解码器进行语音解码,获得语音文本特征向量。
在一些实施例中,所述对文本信息、语音文本特征向量以及图像文本特征向量进行文本预处理,获得短文本数据,包括:提取文本信息、语音文本特征向量以及图像文本特征向量各自的短文本信息;汇集文本信息、语音文本特征向量以及图像文本特征向量各自的短文本信息,获得短文本数据。
在一些实施例中,所述方法还包括:对短文本数据进行数据标注处理,获得待质检数据;在待质检数据满足质检要求的情况下,利用已训练完成的Bert模型对待质检数据进行文本特征提取,获得目标短文本特征数据;利用Focal Loss损失函数对目标短文本特征数据进行短文本分类处理,获得目标短文本数据。
在一些实施例中,按照以下方式训练Bert模型:以基准网络参数对初始Bert模型进行模型训练,获得模型训练后的网络参数以及排序信息;根据模型训练后的网络参数以及排序信息,确定网络参数的目标数量;利用目标数量的网络参数训练初始Bert模型,获得已训练完成的Bert模型。
在一些实施例中,所述利用目标数量的网络参数训练初始Bert模型,获得已训练完成的Bert模型,包括:利用目标数量的网络参数训练初始Bert模型,获得基准Bert模型;利用知识蒸馏技术将教师网络模型迁移至学生网络模型,获得基准学生网络模型。
在一些实施例中,所述利用知识蒸馏技术将教师网络模型迁移至学生网络模型,获得
基准学生网络模型之后,还包括:获得基准学生网络模型的所有编码器的层数以及各编码器的权重;根据所有编码器的层数以及各编码器的权重,确定编码器的目标层数;利用目标层数的编码器训练基准学生网络模型,获得已训练完成的Bert模型。
在一些实施例中,所述装置,包括:语音预处理单元,被配置为对多模态数据相关联的语音信息进行预处理,获得有效语音数据;模型训练单元,被配置为利用第一深度神经网络模型对图像数据进行模型训练,获得图像文本特征向量;语音识别单元,被配置为对有效语音数据进行语音识别处理,获得语音文本特征向量;文本预处理单元,被配置为对文本信息、语音文本特征向量以及图像文本特征向量进行文本预处理,获得短文本数据。
在一些实施例中,所述语音预处理单元还被配置为对音视频信息进行预处理,获得有效视频数据,所述装置还包括:图像帧提取单元,被配置为对有效视频数据进行帧提取处理,获得第二图像数据;图像过滤单元,被配置为对图像信息进行过滤处理,获得第一图像数据;图像融合单元,被配置为对第一图像数据以及第二图像数据进行图像合成处理,获得图像数据。
在一些实施例中,所述装置还包括:数据标注单元,被配置为对短文本数据进行数据标注处理,获得待质检数据;数据质检单元,被配置为在待质检数据满足质检要求的情况下,利用已训练完成的Bert模型对待质检数据进行文本特征提取,获得目标短文本特征数据;文本分类单元,被配置为利用Focal Loss损失函数对目标短文本特征数据进行短文本分类处理,获得目标短文本数据。
在一些实施例中,所述装置包括:处理器和存储有程序指令的存储器,所述处理器被配置为在运行所述程序指令时,执行如前述的用于多模态数据的文本分类方法。
在一些实施例中,所述制冷设备,包括:设备本体;如前述的用于多模态数据的文本分类装置,被安装于所述设备本体。
在一些实施例中,所述存储介质,存储有程序指令,所述程序指令在运行时,执行如前述的用于多模态数据的文本分类方法。
本公开实施例提供的用于多模态数据的文本分类方法、装置、制冷设备和存储介质,可以实现以下技术效果:
本公开实施例根据多模态数据获得文本信息、语音信息和图像信息。再根据语音信息和图像信息,获得语音文本特征向量和图像文本特征向量,以分别利用语音文本特征向量和图像文本特征向量体现出语音和图像各自的文本特征。最后,对文本信息、语音文本特征向量以及图像文本特征向量进行预处理获得短文本数据。如此,通过将同时携带有文本特征的语音文本特征向量和图像文本特征向量与文本信息进行预处理,本公开实施例能够从多模态数据中提取出不同数据来源的短文本特征,从而获得高质量的短文本,进而提升
多模态数据短文本提取的有效性。
以上的总体描述和下文中的描述仅是示例性和解释性的,不用于限制本申请。
一个或多个实施例通过与之对应的附图进行示例性说明,这些示例性说明和附图并不构成对实施例的限定,附图中具有相同参考数字标号的元件示为类似的元件,附图不构成比例限制,并且其中:
图1是本公开实施例提供的用于多模态数据的文本分类装置的示意图;
图2是本公开实施例提供的一个用于多模态数据的文本分类方法的示意图;
图3是本公开实施例提供的另一个用于多模态数据的文本分类方法的示意图;
图4是本公开实施例提供的另一个用于多模态数据的文本分类方法的示意图;
图5是本公开实施例提供的另一个用于多模态数据的文本分类方法的示意图;
图6是本公开实施例提供的另一个用于多模态数据的文本分类方法的示意图;
图7是本公开实施例的一个应用示意图;
图8是本公开实施例提供的一个用于多模态数据的文本分类装置的示意图;
图9是本公开实施例提供的另一个用于多模态数据的文本分类装置的示意图;
图10是本公开实施例提供的一个制冷设备的示意图。
附图标记:
10:语音预处理单元;20:图像过滤单元;30:语音识别单元;40:图像帧提取单元;
50:图像融合单元;60:模型训练单元;70:文本预处理单元;80:数据标注单元;
90:数据质检单元;100:文本分类单元;
200:用于多模态数据的文本分类装置;300:用于多模态数据的文本分类装置;
400:处理器;401:存储器;402:通信接口;403:总线;600:制冷设备。
为了能够更加详尽地了解本公开实施例的特点与技术内容,下面结合附图对本公开实施例的实现进行详细阐述,所附附图仅供参考说明之用,并非用来限定本公开实施例。在以下的技术描述中,为方便解释起见,通过多个细节以提供对所披露实施例的充分理解。然而,在没有这些细节的情况下,一个或多个实施例仍然可以实施。在其它情况下,为简化附图,熟知的结构和装置可以简化展示。
本公开实施例的说明书和权利要求书及上述附图中的术语“第一”、“第二”等是用于区别类似的对象,而不必用于描述特定的顺序或先后次序。应该理解这样使用的数据在适当情况下可以互换,以便这里描述的本公开实施例的实施例。此外,术语“包括”和“具有”以及他们的任何变形,意图在于覆盖不排他的包含。
除非另有说明,术语“多个”表示两个或两个以上。
本公开实施例中,字符“/”表示前后对象是一种“或”的关系。例如,A/B表示:A或B。
术语“和/或”是一种描述对象的关联关系,表示可以存在三种关系。例如,A和/或B,表示:A或B,或,A和B这三种关系。
术语“对应”可以指的是一种关联关系或绑定关系,A与B相对应指的是A与B之间是一种关联关系或绑定关系。
结合图1和图8所示,本公开实施例提供一种用于多模态数据的文本分类装置200,包括语音预处理单元10、模型训练单元60、语音识别单元30以及文本预处理单元70。语音预处理单元10,被配置为对多模态数据相关联的语音信息进行预处理,获得有效语音数据。模型训练单元60,被配置为利用第一深度神经网络模型对图像数据进行模型训练,获得图像文本特征向量。语音识别单元30,被配置为对有效语音数据进行语音识别处理,获得语音文本特征向量。文本预处理单元70,被配置为对文本信息、语音文本特征向量以及图像文本特征向量进行文本预处理,获得短文本数据。
可选地,语音预处理单元10还被配置为对音视频信息进行预处理,获得有效视频数据。用于多模态数据的文本分类系统还包括图像帧提取单元40、图像过滤单元20以及图像融合单元50。图像帧提取单元40,被配置为对有效视频数据进行帧提取处理,获得第二图像数据。图像过滤单元20,被配置为对图像信息进行过滤处理,获得第一图像数据。图像融合单元50,被配置为对第一图像数据以及第二图像数据进行图像合成处理,获得图像数据。
可选地,用于多模态数据的文本分类系统还包括数据标注单元80、数据质检单元90以及文本分类单元100。数据标注单元80,被配置为对短文本数据进行数据标注处理,获得待质检数据。数据质检单元90,被配置为在待质检数据满足质检要求的情况下,利用已训练完成的Bert模型对待质检数据进行文本特征提取,获得目标短文本特征数据。文本分类单元100,被配置为利用Focal Loss损失函数对目标短文本特征数据进行短文本分类处理,获得目标短文本数据。
基于上述用于多模态数据的文本分类系统,结合图2所示,本公开实施例提供一种用于多模态数据的文本分类方法,包括:
S01,处理器根据多模态数据,获得文本信息、语音信息以及图像信息。
该步骤中,多模态数据的类型包括文本信息、语音信息以及图像信息中的部分或者全部。处理器按照以下方式获得多模态数据:处理器通过人机交互模式获取多模态数据。人机交互模式包括但不限于蓝牙、Wi-Fi(Wireless Fidelity,无线保真)、Internet(因特网)、Web(World Wide Web,万维网)、麦克风阵列、拾音器以及软件。其中,软件包括
APP(Application,应用程序)、信息采集工具以及触屏中的一个或者多个。
S02,处理器根据语音信息以及图像信息,获得语音文本特征向量以及图像文本特征向量。
S03,处理器对文本信息、语音文本特征向量以及图像文本特征向量进行文本预处理,获得短文本数据。
采用本公开实施例提供的用于多模态数据的文本分类方法,本公开实施例根据多模态数据获得文本信息、语音信息和图像信息。再根据语音信息和图像信息,获得语音文本特征向量和图像文本特征向量,以分别利用语音文本特征向量和图像文本特征向量体现出语音和图像各自的文本特征。最后,对文本信息、语音文本特征向量以及图像文本特征向量进行预处理获得短文本数据。如此,通过将同时携带有文本特征的语音文本特征向量和图像文本特征向量与文本信息进行预处理,能够从多模态数据中提取出不同数据来源的短文本特征,从而获得高质量的短文本,进而提升多模态数据短文本提取的有效性。
可选地,处理器按照以下方式从多模态数据中提取获得文本信息:处理器从多模态数据中获得初始文本数据。处理器对初始文本数据进行文本预处理,获得文本信息。其中,文本预处理包括分词、词形归一化以及删除停用词。
可选地,结合图3所示,处理器根据语音信息以及图像信息,获得语音文本特征向量以及图像文本特征向量,包括:
S11,处理器对语音信息进行预处理,获得有效语音数据以及有效视频数据。
该步骤中,处理器对语音信息进行预处理,获得有效语音数据以及有效视频数据,包括:
处理器对多模态数据依次进行数据清洗以及格式转换处理,提取有效声音数据。处理器对有效声音数据进行语音视频分离处理,获得语音预处理数据以及有效视频数据。处理器对语音预处理数据进行有效时长处理,获得有效语音数据。如此,通过进行数据清洗以及格式转换处理,能够去除多模态数据中的部分干扰并统一数据格式,获得有效声音数据。通过对有效声音数据进行语音视频分离处理可将语音预处理数据和有效视频数据进行分离,获得语音预处理数据。通过对语音预处理数据进行有效时长处理,能够有效处理或者消除空语音以及背景音等。其中,背景音包括机器运行阶段的机器声音。在实际应用中,处理器可通过数据处理工具或者数据处理脚本对多模态数据进行数据清洗、格式转换处理以及语音视频分离处理。
S12,处理器对有效语音数据进行语音识别处理,获得语音文本特征向量。
该步骤中,处理器对有效语音数据进行语音识别处理,获得语音文本特征向量,包括:处理器输入有效语音数据至第一编码器进行编码处理,获得语音编码数据。处理器利用第
二深度神经网络模型对语音编码数据进行特征提取,获得语音文本特征向量。可选地,第一编码器为Wav2VEC编码器。第二深度神经网络配置有多头注意力机制。如此,通过采用配置多头注意力机制的第二深度神经网络,实现语音编码数据的向量空间的映射,并,提取出语音所涵盖的语义、语法和语境特征,并通过语音文本特征向量予以体现
S13,处理器根据有效视频数据以及图像信息,获得图像文本特征向量。
这样,本公开实施例通过对语音信息进行预处理,可从语音信息中提取出有效语音数据和有效视频数据,降低多模态数据中的干扰并实现数据格式的统一。此外,可有效地消除空语音和背景音。提升有效语音数据和有效视频数据识别的有效性。在此基础上,本公开实施例再对有效语音数据进行语音识别处理,获得语音文本特征向量。有利于从语音中充分挖掘语音文本特征。并根据有效视频数据以及图像信息,获得图像文本特征向量,有利于从图像中充分挖掘图像文本特征。
可选地,结合图4所示,处理器根据有效视频数据以及图像信息,获得图像文本特征向量,包括:
S21,处理器对有效视频数据进行帧提取处理,获得第二图像数据。
S22,处理器对图像信息进行过滤处理,获得第一图像数据。
S23,处理器对第一图像数据以及第二图像数据进行图像合成处理,获得图像数据。
S24,处理器输入图像数据至第一深度神经网络模型进行模型训练,获得图像文本特征向量。
这样,本公开实施例先通过对有效视频数据进行帧提取以从视频数据中提取出第二图像数据,并对图像信息进行过滤处理以从图像信息中提取出第一图像数据。再对第一图像数据和第二图像数据进行图像合成处理以将第一图像数据和第二图像进行有效的融合,生成图像数据。最后,将融合获得的图像数据输入至第一深度神经网络模型进行模型训练,有利于从图像数据中准确地识别语义特征。
可选地,第一深度神经网络模型包括迁移与蒸馏扩散深度融合模型。
其中,迁移与蒸馏扩散深度融合模型表示利用迁移学习,采用预先训练好的复杂模型的输出作为监督信号去训练一简单的网络模型。通常情况下,复杂模型被称为教师模型。简单的网络模型被称为学生模型。
这样,通过选用迁移与蒸馏扩散深度融合模型作为第一深度神经网络模型,可利用该模型的时空以及长距离依赖特性,将教师模型引入时间与空间连续的学生模型,更有利于从图像数据中挖掘出丰富的语义特征,进一步提升图像文本特征向量的有效性。
可选地,处理器对有效语音数据进行语音识别处理,获得语音文本特征向量,包括:
输入有效语音数据至编码器进行语音编码,获得语音编码数据。
输入语音编码数据至解码器进行语音解码,获得语音文本特征向量。
这样,本公开实施例通过输入有效语音数据至编码器进行语音编码,使得编码所获得的语音编码数据不会损失有用语音数据,有利于后续的语音文本特征的提取。
可选地,处理器输入有效语音数据至编码器进行语音编码,获得语音编码数据,包括:处理器利用Wav2VEC编码器对有效语音数据进行编码处理,获得语音编码数据。这样,通过采用Wav2VEC编码器对有效语音数据进行编码,能够增强冗余语音数据的消除效果,从而获得更为纯净的语音编码数据。
可选地,处理器输入语音编码数据至解码器进行语音解码,获得语音文本特征向量,包括:处理器输入语音编码信号至Transformer网络模型进行解码处理,获得语音文本特征向量。这样,利用Transformer网络模型对语音编码信号进行解码处理,能够进一步增强冗余语音数据的消除效果,从而充分提取出语音文本特征。
可选地,处理器对文本信息、语音文本特征向量以及图像文本特征向量进行文本预处理,获得短文本数据,包括:
处理器提取文本信息、语音文本特征向量以及图像文本特征向量各自的短文本信息。
处理器汇集文本信息、语音文本特征向量以及图像文本特征向量各自的短文本信息,获得短文本数据。
这样,本公开实施例通过从文本信息、语音文本特征向量和图像文本特征向量中分别提取出各自的短文本信息,再汇集上述不同数据来源的短文本信息。实现不同数据来源的短文本特征的有效提取,有利于获得高质量的短文本,从而提升多模态数据短文本提取的有效性。
结合图5所示,本公开实施例还提供一种用于多模态数据的文本分类方法,包括:
S31,处理器根据多模态数据,获得文本信息、语音信息以及图像信息。
S32,处理器根据语音信息以及图像信息,获得语音文本特征向量以及图像文本特征向量。
S33,处理器对文本信息、语音文本特征向量以及图像文本特征向量进行文本预处理,获得短文本数据。
S34,处理器对短文本数据进行数据标注处理,获得待质检数据。
该步骤中,处理器对短文本数据进行数据标注处理,获得待质检数据,包括:处理器对短文本数据标注槽位以及元数据属性标签,获得待质检数据。如此,为后续的数据质检提供依据。
S35,处理器在待质检数据满足质检要求的情况下,利用已训练完成的Bert模型对待质检数据进行文本特征提取,获得目标短文本特征数据。
该步骤中,处理器按照以下方式确定待之间数据满足质检要求:元数据属性标签以及食材属性标签正确且错误率小于或者等于2%。
S36,处理器利用Focal Loss损失函数对目标短文本特征进行短文本分类处理,获得目标短文本数据。
采用本公开实施例提供的用于多模态数据的文本分类方法,本公开实施例在获得短文本数据后,依次进行数据标注以及数据质检,并在满足质检要求时,利用已训练完成的Bert模型对待质检数据进行文本特征提取,获得目标短文本特征数据,从而有利于从待质检数据中提取出包含丰富短文本特征的短文本特征数据。由于短文本特征数据存在类型不均衡的问题,影响短文本分类处理的分类准确率。为此,本公开实施例利用Focal Loss损失函数对目标短文本特征进行短文本分类处理,以降低因短文本特征数据的类型不均衡而对短文本分类处理的准确率产生的影响,进一步保证短文本的质量。
可选地,处理器利用Focal Loss损失函数对目标短文本特征进行短文本分类处理,获得目标短文本数据,包括:处理器利用Focal Loss损失函数对目标短文本特征数据进行短文本分类处理,获得短文本分类数据。处理器对短文本分类数据进行数据增强处理,获得目标短文本数据。这样,通过对短文本分类数据进行数据增强处理,能够扩充有限的短文本分类数据集,提升Bert模型的训练效果,增强模型的泛化能力。
可选地,结合图6所示,处理器按照以下方式训练Bert模型:
S41,处理器以基准网络参数对初始Bert模型进行模型训练,获得模型训练后的网络参数以及排序信息。
S42,处理器根据模型训练后的网络参数以及排序信息,确定网络参数的目标数量。
该步骤中,处理器根据模型训练后的网络参数以及排序信息,确定网络参数的目标数量,包括:处理器根据排序信息,选取排序前α的网络参数的数量为目标数量。处理器计算N=α×N0。0.5≤α≤0.8。其中,N0表示网络参数的总数量。N表示网络参数的目标数量。α表示第一权重值。优选的,α=0.75。
S43,处理器利用目标数量的网络参数训练初始Bert模型,获得已训练完成的Bert模型。
这样,针对初始Bert模型进行网络参数的数量的截取,实现对初始Bert模型的微调,保留了模型权重高的网络层数与网络参数,能够减小模型对内存空间的占用。
可选地,处理器按照以下方式获得基准网络参数:
处理器从目标短文本数据中提取获得训练数据。
处理器利用训练数据对初始Bert模型进行模型训练,获得基准网络参数。
这样,有利于提升基准网络参数的准确性,提升Bert模型的训练效果。
可选地,处理器利用目标数量的网络参数训练初始Bert模型,获得已训练完成的Bert模型,包括:
处理器利用目标数量的网络参数训练初始Bert模型,获得基准Bert模型。处理器利用知识蒸馏技术将教师网络模型迁移至学生网络模型,获得基准学生网络模型。其中,基准Bert模型为教师网络模型。基准学生网络模型为已训练完成的Bert模型。如此,通过教师网络模型至学生网络模型的迁移,以利用大模型指导小模型进行模型训练,所获得的基准学生网络模型具有与教师网络模型相当的性能,同时,大幅降低网络参数的数量,实现模型的压缩与加速。
可选地,处理器利用知识蒸馏技术将教师网络模型迁移至学生网络模型,获得基准学生网络模型之后,还包括:处理器获得基准学生网络模型的所有编码器的层数以及各编码器的权重。处理器根据所有编码器的层数以及各编码器的权重,确定编码器的目标层数。处理器利用目标层数的编码器训练基准学生网络模型,获得已训练完成的Bert模型。
这样,经过裁剪后的Bert模型的编码器层数有所降低,能减小模型对内存空间的占用,提高Bert模型的模型训练的效率。
可选地,处理器根据所有编码器的层数以及各编码器的权重,确定编码器的目标层数,包括:处理器按照权重大小对所有编码器的权重进行降序排列。处理器从降序排列的权重中,选取排序前β位的编码器的层数为目标层数。处理器计算M=β×M0。0.5≤β≤0.8。其中,M0表示所有编码器的层数。M表示编码器的目标层数。β表示第二权重值。优选的,β=0.75。
在实际应用中,如图7所示,用于多模态数据的文本分类方法包括以下步骤:
S101:处理器采集多模态数据。
S102:处理器对多模态数据进行数据清洗,获得语音信息、文本信息和图像信息。
S103:处理器对语音信息进行进行预处理,获得有效语音数据以及有效视频数据。
S1041:处理器输入有效语音数据依次经Wav2编码器、Transformer解码器,获得语音文本特征向量。
S1042:处理器对有效视频数据进行图像帧提取,获得第二目标图像。
S1043:处理器对图像信息进行图像过滤,获得第一目标图像。
S1044:处理器对第二目标图像和第一目标图像进行图像合成,获得图像数据。
S1045:处理器输入图像数据至迁移与蒸馏扩散深度融合模型,进行向量空间的映射处理,以从图像数据中提取出来图像文本特征向量。
S105:处理器对文本信息、语音文本特征向量以及图像文本特征向量进行文本预处理,获得短文本数据。
S106:处理器对短文本数据进行数据标注处理,获得待质检数据。
S107:处理器判断待质检数据是否满足质检要求,若是,则执行S108,否则,执行S109。
S108:处理器将待质检数据作为文本分类结果进行输出。
S109:处理器利用已训练完成的Bert模型对待质检数据进行文本特征提取,获得目标短文本特征数据。
S110:处理器利用Focal Loss损失函数对目标短文本特征进行短文本分类处理,获得目标短文本数据,并将目标短文本数据作为文本分类结果进行输出。
结合图9所示,本公开实施例提供一种用于多模态数据的文本分类装置300,包括处理器(processor)400和存储器(memory)401。可选地,该装置还可以包括通信接口(Communication Interface)402和总线403。其中,处理器400、通信接口402、存储器401可以通过总线403完成相互间的通信。通信接口402可以用于信息传输。处理器400可以调用存储器401中的逻辑指令,以执行上述实施例的用于多模态数据的文本分类方法。
此外,上述的存储器401中的逻辑指令可以通过软件功能单元的形式实现并作为独立的产品销售或使用时,可以存储在一个计算机可读取存储介质中。
存储器401作为一种计算机可读存储介质,可用于存储软件程序、计算机可执行程序,如本公开实施例中的方法对应的程序指令/模块。处理器400通过运行存储在存储器401中的程序指令/模块,从而执行功能应用以及数据处理,即实现上述实施例中用于多模态数据的文本分类方法。
存储器401可包括存储程序区和存储数据区,其中,存储程序区可存储操作系统、至少一个功能所需的应用程序;存储数据区可存储根据终端设备的使用所创建的数据等。此外,存储器401可以包括高速随机存取存储器,还可以包括非易失性存储器。
结合图10所示,本公开实施例提供了一种制冷设备600,包括:设备本体,以及上述的用于多模态数据的文本分类装置200(300)。用于多模态数据的文本分类装置200(300)被安装于设备本体。这里所表述的安装关系,并不仅限于在产品内部放置,还包括了与产品的其他元器件的安装连接,包括但不限于物理连接、电性连接或者信号传输连接等。本领域技术人员可以理解的是,用于多模态数据的文本分类装置200(300)可以适配于可行的产品主体,进而实现其他可行的实施例。
本公开实施例提供了一种计算机可读存储介质,存储有计算机可执行指令,所述计算机可执行指令设置为执行上述用于多模态数据的文本分类方法。
本公开实施例提供了一种计算机程序,当所述计算机程序被计算机执行时,使所述计算机实现上述用于多模态数据的文本分类方法。
本公开实施例提供了一种计算机程序产品,所述计算机程序产品包括存储在计算机可
读存储介质上的计算机指令,当所述程序指令被计算机执行时,使所述计算机实现上述用于多模态数据的文本分类方法。
上述的计算机可读存储介质可以是暂态计算机可读存储介质,也可以是非暂态计算机可读存储介质。
本公开实施例的技术方案可以以软件产品的形式体现出来,该计算机软件产品存储在一个存储介质中,包括一个或多个指令用以使得一台计算机设备(可以是个人计算机,服务器,或者网络设备等)执行本公开实施例所述方法的全部或部分步骤。而前述的存储介质可以是非暂态存储介质,包括:U盘、移动硬盘、只读存储器(ROM,Read-Only Memory)、随机存取存储器(RAM,Random Access Memory)、磁碟或者光盘等多种可以存储程序代码的介质,也可以是暂态存储介质。
说明书中对任何现有技术的引用不是也不应被视为承认或以任何形式暗示该现有技术构成申请地区或任何其他司法管辖区的公知常识的一部分,或者该现有技术可以被本领域技术人员合理地理解和视为相关。
以上描述和附图充分地示出了本公开的实施例,以使本领域的技术人员能够实践它们。其他实施例可以包括结构的、逻辑的、电气的、过程的以及其他的改变。实施例仅代表可能的变化。除非明确要求,否则单独的部件和功能是可选的,并且操作的顺序可以变化。一些实施例的部分和特征可以被包括在或替换其他实施例的部分和特征。而且,本申请中使用的用词仅用于描述实施例并且不用于限制权利要求。如在实施例以及权利要求的描述中使用的,除非上下文清楚地表明,否则单数形式的“一个”(a)、“一个”(an)和“所述”(the)旨在同样包括复数形式。类似地,如在本申请中所使用的术语“和/或”是指包含一个或一个以上相关联的列出的任何以及所有可能的组合。另外,当用于本申请中时,术语“包括”(comprise)及其变型“包括”(comprises)和/或包括(comprising)等指陈述的特征、整体、步骤、操作、元素,和/或组件的存在,但不排除一个或一个以上其它特征、整体、步骤、操作、元素、组件和/或这些的分组的存在或添加。在没有更多限制的情况下,由语句“包括一个…”限定的要素,并不排除在包括所述要素的过程、方法或者设备中还存在另外的相同要素。本文中,每个实施例重点说明的可以是与其他实施例的不同之处,各个实施例之间相同相似部分可以互相参见。对于实施例公开的方法、产品等而言,如果其与实施例公开的方法部分相对应,那么相关之处可以参见方法部分的描述。
本领域技术人员可以意识到,结合本文中所公开的实施例描述的各示例的单元及算法步骤,能够以电子硬件、或者计算机软件和电子硬件的结合来实现。这些功能究竟以硬件还是软件方式来执行,可以取决于技术方案的特定应用和设计约束条件。所述技术人员可以对每个特定的应用来使用不同方法以实现所描述的功能,但是这种实现不应认为超出本
公开实施例的范围。所述技术人员可以清楚地了解到,为描述的方便和简洁,上述描述的系统、装置和单元的具体工作过程,可以参考前述方法实施例中的对应过程,在此不再赘述。
本文所披露的实施例中,所揭露的方法、产品(包括但不限于装置、设备等),可以通过其它的方式实现。例如,以上所描述的装置实施例仅仅是示意性的,例如,所述单元的划分,可以仅仅为一种逻辑功能划分,实际实现时可以有另外的划分方式,例如多个单元或组件可以结合或者可以集成到另一个系统,或一些特征可以忽略,或不执行。另外,所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口,装置或单元的间接耦合或通信连接,可以是电性,机械或其它的形式。所述作为分离部件说明的单元可以是或者也可以不是物理上分开的,作为单元显示的部件可以是或者也可以不是物理单元,即可以位于一个地方,或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例。另外,在本公开实施例中的各功能单元可以集成在一个处理单元中,也可以是各个单元单独物理存在,也可以两个或两个以上单元集成在一个单元中。
附图中的流程图和框图显示了根据本公开实施例的系统、方法和计算机程序产品的可能实现的体系架构、功能和操作。在这点上,流程图或框图中的每个方框可以代表一个模块、程序段或代码的一部分,所述模块、程序段或代码的一部分包含一个或多个用于实现规定的逻辑功能的可执行指令。在有些作为替换的实现中,方框中所标注的功能也可以以不同于附图中所标注的顺序发生。例如,两个连续的方框实际上可以基本并行地执行,它们有时也可以按相反的顺序执行,这可以依所涉及的功能而定。在附图中的流程图和框图所对应的描述中,不同的方框所对应的操作或步骤也可以以不同于描述中所披露的顺序发生,有时不同的操作或步骤之间不存在特定的顺序。例如,两个连续的操作或步骤实际上可以基本并行地执行,它们有时也可以按相反的顺序执行,这可以依所涉及的功能而定。框图和/或流程图中的每个方框、以及框图和/或流程图中的方框的组合,可以用执行规定的功能或动作的专用的基于硬件的系统来实现,或者可以用专用硬件与计算机指令的组合来实现。
Claims (18)
- 一种用于多模态数据的文本分类方法,其特征在于,包括:根据多模态数据,获得文本信息、语音信息以及图像信息;根据语音信息以及图像信息,获得语音文本特征向量以及图像文本特征向量;对文本信息、语音文本特征向量以及图像文本特征向量进行文本预处理,获得短文本数据。
- 根据权利要求1所述的方法,其中,所述根据语音信息以及图像信息,获得语音文本特征向量以及图像文本特征向量,包括:对语音信息进行预处理,获得有效语音数据以及有效视频数据;对有效语音数据进行语音识别处理,获得语音文本特征向量;根据有效视频数据以及图像信息,获得图像文本特征向量。
- 根据权利要求2所述的方法,其中,所述根据有效视频数据以及图像信息,获得图像文本特征向量,包括:对有效视频数据进行帧提取处理,获得第二图像数据;对图像信息进行过滤处理,获得第一图像数据;对第一图像数据以及第二图像数据进行图像合成处理,获得图像数据;输入图像数据至第一深度神经网络模型进行模型训练,获得图像文本特征向量。
- 根据权利要求3所述的方法,其中,所述第一深度神经网络模型包括迁移与蒸馏扩散深度融合模型。
- 根据权利要求2所述的方法,其中,所述对有效语音数据进行语音识别处理,获得语音文本特征向量,包括:输入有效语音数据至编码器进行语音编码,获得语音编码数据;输入语音编码数据至解码器进行语音解码,获得语音文本特征向量。
- 根据权利要求1至5任一项所述的方法,其中,所述对文本信息、语音文本特征向量以及图像文本特征向量进行文本预处理,获得短文本数据,包括:提取文本信息、语音文本特征向量以及图像文本特征向量各自的短文本信息;汇集文本信息、语音文本特征向量以及图像文本特征向量各自的短文本信息,获得短文本数据。
- 根据权利要求1至6任一项所述的方法,其中,还包括:对短文本数据进行数据标注处理,获得待质检数据;在待质检数据满足质检要求的情况下,利用已训练完成的Bert模型对待质检数据 进行文本特征提取,获得目标短文本特征数据;利用Focal Loss损失函数对目标短文本特征数据进行短文本分类处理,获得目标短文本数据。
- 根据权利要求7所述的方法,其中,按照以下方式训练Bert模型:以基准网络参数对初始Bert模型进行模型训练,获得模型训练后的网络参数以及排序信息;根据模型训练后的网络参数以及排序信息,确定网络参数的目标数量;利用目标数量的网络参数训练初始Bert模型,获得已训练完成的Bert模型。
- 根据权利要求8所述的方法,其中,所述利用目标数量的网络参数训练初始Bert模型,获得已训练完成的Bert模型,包括:利用目标数量的网络参数训练初始Bert模型,获得基准Bert模型;利用知识蒸馏技术将教师网络模型迁移至学生网络模型,获得基准学生网络模型。
- 根据权利要求9所述的方法,其中,所述利用知识蒸馏技术将教师网络模型迁移至学生网络模型,获得基准学生网络模型之后,还包括:获得基准学生网络模型的所有编码器的层数以及各编码器的权重;根据所有编码器的层数以及各编码器的权重,确定编码器的目标层数;利用目标层数的编码器训练基准学生网络模型,获得已训练完成的Bert模型。
- 一种用于多模态数据的文本分类装置,其特征在于,包括:语音预处理单元,被配置为对多模态数据相关联的语音信息进行预处理,获得有效语音数据;模型训练单元,被配置为利用第一深度神经网络模型对图像数据进行模型训练,获得图像文本特征向量;语音识别单元,被配置为对有效语音数据进行语音识别处理,获得语音文本特征向量;文本预处理单元,被配置为对文本信息、语音文本特征向量以及图像文本特征向量进行文本预处理,获得短文本数据。
- 根据权利要求11所述的装置,其中,所述语音预处理单元还被配置为对音视频信息进行预处理,获得有效视频数据,所述装置还包括:图像帧提取单元,被配置为对有效视频数据进行帧提取处理,获得第二图像数据;图像过滤单元,被配置为对图像信息进行过滤处理,获得第一图像数据;图像融合单元,被配置为对第一图像数据以及第二图像数据进行图像合成处理, 获得图像数据。
- 根据权利要求11所述的装置,其中,所述装置还包括:数据标注单元,被配置为对短文本数据进行数据标注处理,获得待质检数据;数据质检单元,被配置为在待质检数据满足质检要求的情况下,利用已训练完成的Bert模型对待质检数据进行文本特征提取,获得目标短文本特征数据;文本分类单元,被配置为利用Focal Loss损失函数对目标短文本特征数据进行短文本分类处理,获得目标短文本数据。
- 一种用于多模态数据的文本分类装置,包括处理器和存储有程序指令的存储器,其特征在于,所述处理器被配置为在运行所述程序指令时,执行如权利要求1至10任一项所述的用于多模态数据的文本分类方法。
- 一种制冷设备,其特征在于,包括:设备本体;如权利要求11至14任一项所述的用于多模态数据文本分类装置,被安装于所述设备本体。
- 一种存储介质,存储有程序指令,其特征在于,所述程序指令在运行时,执行如权利要求1至10任一项所述的用于多模态数据的文本分类方法。
- 一种计算机程序,当所述计算机程序被计算机执行时,使所述计算机实现如权利要求1至10任一项所述的用于多模态数据的文本分类方法。
- 一种计算机程序产品,所述计算机程序产品包括存储在计算机可读存储介质上的计算机指令,当所述程序指令被计算机执行时,使所述计算机实现如权利要求1至10任一项所述的用于多模态数据的文本分类方法。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310694668.X | 2023-06-12 | ||
| CN202310694668.XA CN116910244A (zh) | 2023-06-12 | 2023-06-12 | 用于多模态数据的文本分类方法及装置、制冷设备、介质 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2024255723A1 true WO2024255723A1 (zh) | 2024-12-19 |
Family
ID=88363754
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2024/098340 Ceased WO2024255723A1 (zh) | 2023-06-12 | 2024-06-11 | 用于多模态数据的文本分类方法及装置、制冷设备、介质 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN116910244A (zh) |
| WO (1) | WO2024255723A1 (zh) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119693634A (zh) * | 2025-02-24 | 2025-03-25 | 浙江大华技术股份有限公司 | 目标检测方法、设备及存储介质 |
| CN119990782A (zh) * | 2025-04-11 | 2025-05-13 | 杭州市公安局高新技术产业开发区分局、杭州市公安局滨江区分局 | 企业经营风险预警方法、装置、计算机设备和介质 |
| CN120726420A (zh) * | 2025-08-14 | 2025-09-30 | 武汉工程大学 | 一种基于图像的五元特征检测模型训练方法、系统及设备 |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116910244A (zh) * | 2023-06-12 | 2023-10-20 | 青岛海尔电冰箱有限公司 | 用于多模态数据的文本分类方法及装置、制冷设备、介质 |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111079623A (zh) * | 2019-12-09 | 2020-04-28 | 成都睿沿科技有限公司 | 一种目标检测方法、装置及存储介质 |
| US20220327809A1 (en) * | 2021-07-12 | 2022-10-13 | Beijing Baidu Netcom Science Technology Co., Ltd. | Method, device and storage medium for training model based on multi-modal data joint learning |
| CN115269774A (zh) * | 2022-06-30 | 2022-11-01 | 青岛海尔科技有限公司 | 文本意图的识别方法和装置、存储介质和电子装置 |
| CN116108176A (zh) * | 2022-12-31 | 2023-05-12 | 青岛海尔电冰箱有限公司 | 基于多模态深度学习的文本分类方法、设备及存储介质 |
| CN116186258A (zh) * | 2022-12-31 | 2023-05-30 | 青岛海尔电冰箱有限公司 | 基于多模态知识图谱的文本分类方法、设备及存储介质 |
| CN116910244A (zh) * | 2023-06-12 | 2023-10-20 | 青岛海尔电冰箱有限公司 | 用于多模态数据的文本分类方法及装置、制冷设备、介质 |
| CN116956209A (zh) * | 2023-06-12 | 2023-10-27 | 青岛海尔电冰箱有限公司 | 用于多模态数据的文本提取方法及装置、制冷设备、介质 |
-
2023
- 2023-06-12 CN CN202310694668.XA patent/CN116910244A/zh active Pending
-
2024
- 2024-06-11 WO PCT/CN2024/098340 patent/WO2024255723A1/zh not_active Ceased
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111079623A (zh) * | 2019-12-09 | 2020-04-28 | 成都睿沿科技有限公司 | 一种目标检测方法、装置及存储介质 |
| US20220327809A1 (en) * | 2021-07-12 | 2022-10-13 | Beijing Baidu Netcom Science Technology Co., Ltd. | Method, device and storage medium for training model based on multi-modal data joint learning |
| CN115269774A (zh) * | 2022-06-30 | 2022-11-01 | 青岛海尔科技有限公司 | 文本意图的识别方法和装置、存储介质和电子装置 |
| CN116108176A (zh) * | 2022-12-31 | 2023-05-12 | 青岛海尔电冰箱有限公司 | 基于多模态深度学习的文本分类方法、设备及存储介质 |
| CN116186258A (zh) * | 2022-12-31 | 2023-05-30 | 青岛海尔电冰箱有限公司 | 基于多模态知识图谱的文本分类方法、设备及存储介质 |
| CN116910244A (zh) * | 2023-06-12 | 2023-10-20 | 青岛海尔电冰箱有限公司 | 用于多模态数据的文本分类方法及装置、制冷设备、介质 |
| CN116956209A (zh) * | 2023-06-12 | 2023-10-27 | 青岛海尔电冰箱有限公司 | 用于多模态数据的文本提取方法及装置、制冷设备、介质 |
Non-Patent Citations (1)
| Title |
|---|
| HUAN LIU, ZHANG ZHIXIONG , WANG YUFEI: "A Review on Main Optimization Methods of BERT", DATA ANALYSIS AND KNOWLEDGE DISCOVERY, 4 November 2020 (2020-11-04), pages 1 - 18, XP055935777, ISSN: 2096-3467, DOI: 10.11925/infotech.2096-3467.2020.0965 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119693634A (zh) * | 2025-02-24 | 2025-03-25 | 浙江大华技术股份有限公司 | 目标检测方法、设备及存储介质 |
| CN119990782A (zh) * | 2025-04-11 | 2025-05-13 | 杭州市公安局高新技术产业开发区分局、杭州市公安局滨江区分局 | 企业经营风险预警方法、装置、计算机设备和介质 |
| CN120726420A (zh) * | 2025-08-14 | 2025-09-30 | 武汉工程大学 | 一种基于图像的五元特征检测模型训练方法、系统及设备 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116910244A (zh) | 2023-10-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2024255723A1 (zh) | 用于多模态数据的文本分类方法及装置、制冷设备、介质 | |
| US20240212706A1 (en) | Audio data processing | |
| Khan et al. | MemoCMT: multimodal emotion recognition using cross-modal transformer-based feature fusion | |
| KR102660124B1 (ko) | 동영상 감정 학습용 데이터 생성 방법, 동영상 감정 판단 방법, 이를 이용하는 동영상 감정 판단 장치 | |
| CN110909613B (zh) | 视频人物识别方法、装置、存储介质与电子设备 | |
| WO2024140434A1 (zh) | 基于多模态知识图谱的文本分类方法、设备及存储介质 | |
| CN107220235B (zh) | 基于人工智能的语音识别纠错方法、装置及存储介质 | |
| US10339452B2 (en) | Automated ontology development | |
| CN104735468B (zh) | 一种基于语义分析将图像合成新视频的方法及系统 | |
| Doumbouya et al. | Using radio archives for low-resource speech recognition: towards an intelligent virtual assistant for illiterate users | |
| WO2024140430A9 (zh) | 基于多模态深度学习的文本分类方法、设备及存储介质 | |
| JP4109185B2 (ja) | 映像シーン区間情報抽出方法,映像シーン区間情報抽出装置,映像シーン区間情報抽出プログラムおよびそのプログラムを記録した記録媒体 | |
| CN116682411A (zh) | 语音合成方法、语音合成系统、电子设备及存储介质 | |
| CN116956209A (zh) | 用于多模态数据的文本提取方法及装置、制冷设备、介质 | |
| EP4708285A1 (en) | Model determination method, model application method, and related device | |
| CN113312456A (zh) | 短视频文本生成方法、装置、设备及存储介质 | |
| KR102069101B1 (ko) | 고객의 소리 데이터로부터 주요 자질 추출 방법 및 이를 이용한 데이터 유형 분류 방법 | |
| KR102435243B1 (ko) | 음성 정보로부터 변환된 텍스트의 영상 리소스 매칭을 이용한 멀티미디어 변환 콘텐츠 제작 서비스 제공 방법 | |
| CN117290524A (zh) | 用于信息推荐的方法、装置和家电设备 | |
| Tsalera et al. | CNN-based Segmentation and Classification of Sound Streams under realistic conditions | |
| Li et al. | Cflip: Generalizing lipreading to unseen speakers by learning common features | |
| CN116543780A (zh) | 模型更新方法和装置、语音转换方法、设备及存储介质 | |
| CN113626637A (zh) | 视频数据筛选法、装置、计算机设备和存储介质 | |
| CN113360642A (zh) | 文本数据处理方法及装置、存储介质和电子设备 | |
| CN114373448B (zh) | 话题检测方法、装置、电子设备及存储介质 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 24822673 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |