CN114565804A - NLP model training and recognizing system - Google Patents

NLP model training and recognizing system Download PDF

Info

Publication number
CN114565804A
CN114565804A CN202210154146.6A CN202210154146A CN114565804A CN 114565804 A CN114565804 A CN 114565804A CN 202210154146 A CN202210154146 A CN 202210154146A CN 114565804 A CN114565804 A CN 114565804A
Authority
CN
China
Prior art keywords
training
model
unit
picture
text
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.)
Withdrawn
Application number
CN202210154146.6A
Other languages
Chinese (zh)
Inventor
盛夏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Individual
Original Assignee
Individual
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Individual filed Critical Individual
Priority to CN202210154146.6A priority Critical patent/CN114565804A/en
Publication of CN114565804A publication Critical patent/CN114565804A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR 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
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR 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; CALCULATING OR 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses an NLP model training and identifying system which comprises a training updating module and a training image obtaining unit, wherein the training image obtaining unit is used for obtaining sample pictures and description information of each sample picture; the training updating module is used for training and extracting corresponding characteristics by utilizing the sample pictures and the description information of the training image acquisition unit, inputting the corresponding characteristics into the established model to obtain a fusion characteristic vector output by the model, and retrieving pictures based on the fusion characteristic vector; the training updating module comprises a training set, a feature extraction unit, a first training unit, a calculation analysis unit, an updating unit and a second training unit. The invention belongs to the technical field of learning and training systems, and particularly provides an NLP model training and recognition system for realizing more accurate picture retrieval by starting from extracting picture features and related character information.

Description

NLP model training and recognizing system
Technical Field
The invention belongs to the technical field of learning and training systems, particularly relates to the technical fields of deep learning, NLP (non line segment prediction) and the like, and particularly relates to an NLP model training and recognizing system.
Background
NLP (Natural Language processing) is a sub-field of Artificial Intelligence (AI), at present, a conventional NLP needs three processes of model training, model issuing and model starting before being capable of providing normal interface service, the three processes all need to be manually operated on a server by a developer in the industry, misoperation risks exist during issuing and training of a NLP semantic recognition model each time, and faults are easily caused by human operation errors.
With the continuous progress of basic technology, the human information interaction form is continuously evolved, from sound, characters and pictures to video, the form is more vivid and efficient, but is also more huge and complex. Among them, picture media have recently become a common interactive form, and efficient retrieval of pictures has been widely applied in fields of search, recommendation, advertisement, and the like, and is also a hot spot of research in the industry.
Disclosure of Invention
In view of the above situation, in order to overcome the defects in the prior art, the present invention provides an NLP model training and recognition system which starts from extracting picture features and related text information and realizes more accurate picture retrieval.
The technical scheme adopted by the invention is as follows: the invention relates to an NLP model training and identifying system, which comprises a training updating module and a training image acquisition unit,
the training image acquisition unit is used for acquiring sample pictures and description information of each sample picture;
the training updating module utilizes the sample picture and the description information of the training image obtaining unit to train and extract corresponding characteristics, inputs the corresponding characteristics into the established model to obtain a fusion characteristic vector output by the model, and carries out picture retrieval based on the fusion characteristic vector.
In the scheme, the training updating module comprises a training set, a characteristic extracting unit, a first training unit, a calculation analyzing unit, an updating unit and a second training unit,
the characteristic extraction unit is used for extracting picture characteristics from a sample picture to be retrieved and extracting text characteristics from the sample picture, the picture characteristics and the text characteristics extracted by the characteristic extraction unit are stored in a training set,
the first training unit is used for respectively training the generator and the discriminator of the image translation model by using the picture characteristics and the text characteristic data in the training set and determining the trained image translation model as a first model;
a computational analysis unit configured to input a current training data sequence into the first model, calculate a first gradient of a meta-loss function over the current training data sequence, and perform text analysis on the training data using Natural Language Processing (NLP);
an updating unit configured to update the network parameters of the first model according to the first gradient and a learning rate, resulting in a second model;
and the second training unit is configured to perform model training based on the future training data sequence and the updated learning rate, and form a target model after the training is completed on the basis of a second model.
Preferably, the description information of the sample picture includes, but is not limited to, text information on the sample picture, and also includes text information associated with the sample picture.
Preferably, the calculation and analysis unit further includes a text merging module, a keyword extraction module, and a text analysis unit, the text merging module is configured to merge text information on the sample picture and text information associated with the target picture, the keyword extraction module is configured to form text analysis on the text information of the target picture to obtain a set of vectors composed of keywords and weights, and the text analysis unit is configured to perform keyword extraction and text classification to obtain vectors and classification.
Preferably, the generator in the first training unit is a coding model-decoding model structure, and the coding model adopts a residual network architecture.
The invention with the structure has the following beneficial effects: the scheme is an NLP model training and recognizing system, and aims to solve the problem that in the prior art, retrieval understanding of pictures surrounds picture pixels, and more accurate picture retrieval is achieved by extracting picture features and relevant character information.
Drawings
Fig. 1 is a schematic diagram of the overall composition of an NLP model training recognition system according to the present invention.
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments; all other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1, the NLP model training and recognition system of the present invention includes a training update module and a training image acquisition unit.
The training image acquisition unit is used for acquiring sample pictures and description information of each sample picture, the training updating module is used for training and extracting corresponding features by using the sample pictures and the description information of the training image acquisition unit, inputting the corresponding features into the established model to obtain a fusion feature vector output by the model, and searching pictures based on the fusion feature vector.
In the scheme, the training updating module comprises a training set, a characteristic extracting unit, a first training unit, a calculation analyzing unit, an updating unit and a second training unit,
the characteristic extraction unit is used for extracting picture characteristics of a target picture to be retrieved and extracting text characteristics of the target picture, and the picture characteristics and the text characteristics extracted by the characteristic extraction unit are stored in a training set;
the first training unit is used for respectively training the generator and the discriminator of the image translation model by using the picture characteristics and the text characteristic data in the training set and determining the trained image translation model as a first model;
a computational analysis unit configured to input a current training data sequence into the first model, calculate a first gradient of a meta-loss function over the current training data sequence, and perform text analysis on the training data using Natural Language Processing (NLP);
an updating unit configured to update the network parameters of the first model according to the first gradient and a learning rate, resulting in a second model;
and the second training unit is configured to perform model training based on the future training data sequence and the updated learning rate, and a target model is formed after the training is completed.
In this scheme, the description information of the sample picture includes, but is not limited to, text information on the sample picture, and also includes text information associated with the sample picture.
As a preferred scheme, the calculation and analysis unit further comprises a text merging module, a keyword extraction module and a text analysis unit, wherein the text merging module is used for merging the text information on the target picture and the text information associated with the target picture, the keyword extraction module is used for performing text analysis on the text information of the target picture to obtain a group of vectors consisting of keywords and weights, and the text analysis unit is used for performing keyword extraction and text classification to obtain the vectors and classification.
Preferably, the generator in the first training unit is in a coding model-decoding model structure, and the coding model adopts a residual error network architecture.
At present, in the prior art, a generator G of an image translation model adopts a coding model-decoding model, that is, an encoder-decoder model, wherein the coding model and the decoding model can adopt any one of deep learning algorithms such as CNN, RNN, BiRNN, LSTM, etc., and when a network is deep, the model effect is increasingly poor in the current deep learning algorithm, and experiments can find that: with the continuous increase of network levels, the precision of a model is continuously improved, and after the network levels are increased to a certain number, the training precision and the testing precision are rapidly reduced, which indicates that a deep network becomes more difficult to train after the network becomes very deep, so in order to reduce errors, the precision of the model is maintained by adopting a residual network architecture (RestNetBlock), and the precision of the model can be maintained when the network levels are very deep through a residual network structure.
In the embodiment of the disclosure, in the case that the picture retrieval based on the target picture is required, the text description information of the target picture can be acquired from one or more dimensions.
In embodiment 1, the text information manually labeled for the target picture is acquired by the training image acquisition unit, and the text information of the target picture can be acquired according to the source of the target picture. For example, for a target picture under a certain term from the encyclopedia, the text information under the term in the encyclopedia can be used as the text information of the target picture.
The text combining module is used for combining the character information on the target picture and the character information related to the target picture, the keyword extracting module is used for forming text analysis on the character information of the target picture to obtain a group of vectors consisting of keywords and weights, and the text analyzing unit is used for extracting the keywords and classifying the texts to obtain the vectors and the classifications.
After the target picture and the text description information of the target picture are obtained, picture features can be extracted from the target picture, text features are extracted from the text description information of the target picture, and then the extracted picture features and the extracted text features are input into a pre-trained target model, so that a fusion feature vector is obtained. And finally, searching the picture based on the fusion feature vector, so that a relatively accurate searching result can be obtained.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
The present invention and its embodiments have been described above, and the description is not intended to be limiting, and the drawings are only one embodiment of the present invention, and the actual structure is not limited thereto. In summary, those skilled in the art should appreciate that they can readily use the disclosed conception and specific embodiments as a basis for designing or modifying other structures for carrying out the same purposes of the present invention without departing from the spirit and scope of the invention as defined by the appended claims.

Claims (5)

1. The utility model provides a NLP model training identification system which characterized in that: the system comprises a training updating module and a training image acquisition unit, wherein the training image acquisition unit is used for acquiring sample pictures and description information of each sample picture; the training updating module utilizes the sample picture and the description information of the training image obtaining unit to train and extract corresponding characteristics, inputs the corresponding characteristics into the established model to obtain a fusion characteristic vector output by the model, and carries out picture retrieval based on the fusion characteristic vector.
2. The NLP model training and recognition system of claim 1, wherein: the training updating module comprises a training set, a feature extraction unit, a first training unit, a calculation analysis unit, an updating unit and a second training unit;
the characteristic extraction unit is used for extracting picture characteristics from a sample picture to be retrieved and extracting text characteristics from the sample picture, and the picture characteristics and the text characteristics extracted by the characteristic extraction unit are stored in a training set;
the first training unit is used for respectively training a generator and a discriminator of the image translation model by using the picture characteristics and the text characteristic data in the training set and determining the trained image translation model as a first model;
the computational analysis unit is configured to input a current training data sequence into the first model, calculate a first gradient of a meta-loss function on the current training data sequence, and perform text analysis on the training data using Natural Language Processing (NLP);
the updating unit is configured to update the network parameters of the first model according to the first gradient and the learning rate to obtain a second model;
the second training unit is configured to perform model training based on the future training data sequence and the updated learning rate, and form a target model after the training is completed.
3. The NLP model training and recognition system of claim 2, wherein: the description information of the sample picture includes, but is not limited to, text information on the sample picture, and also includes text information associated with the sample picture.
4. The NLP model training and recognition system of claim 3, wherein: the calculation analysis unit further comprises a text merging module, a keyword extraction module and a text analysis unit, wherein the text merging module is used for merging the character information on the target picture and the character information related to the target picture, the keyword extraction module is used for performing text analysis on the character information of the target picture to obtain a group of vectors consisting of keywords and weights, and the text analysis unit is used for performing keyword extraction and text classification to obtain vectors and classification.
5. The NLP model training and recognition system of claim 2, wherein: the generator is in a coding model-decoding model structure, and the coding model adopts a residual error network architecture.
CN202210154146.6A 2022-02-21 2022-02-21 NLP model training and recognizing system Withdrawn CN114565804A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210154146.6A CN114565804A (en) 2022-02-21 2022-02-21 NLP model training and recognizing system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210154146.6A CN114565804A (en) 2022-02-21 2022-02-21 NLP model training and recognizing system

Publications (1)

Publication Number Publication Date
CN114565804A true CN114565804A (en) 2022-05-31

Family

ID=81714226

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210154146.6A Withdrawn CN114565804A (en) 2022-02-21 2022-02-21 NLP model training and recognizing system

Country Status (1)

Country Link
CN (1) CN114565804A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115205635A (en) * 2022-09-13 2022-10-18 有米科技股份有限公司 Weak supervision self-training method and device of image-text semantic alignment model

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115205635A (en) * 2022-09-13 2022-10-18 有米科技股份有限公司 Weak supervision self-training method and device of image-text semantic alignment model

Similar Documents

Publication Publication Date Title
CN108829822B (en) Media content recommendation method and device, storage medium and electronic device
CN106845411B (en) Video description generation method based on deep learning and probability map model
CN111444340A (en) Text classification and recommendation method, device, equipment and storage medium
CN109165563B (en) Pedestrian re-identification method and apparatus, electronic device, storage medium, and program product
CN108765383A (en) Video presentation method based on depth migration study
CN113128431B (en) Video clip retrieval method, device, medium and electronic equipment
CN111831924A (en) Content recommendation method, device, equipment and readable storage medium
CN114691864A (en) Text classification model training method and device and text classification method and device
CN113392265A (en) Multimedia processing method, device and equipment
CN110298046B (en) Translation model training method, text translation method and related device
CN115408488A (en) Segmentation method and system for novel scene text
CN113408287B (en) Entity identification method and device, electronic equipment and storage medium
CN114661951A (en) Video processing method and device, computer equipment and storage medium
CN114565804A (en) NLP model training and recognizing system
CN117332789A (en) Semantic analysis method and system for dialogue scene
CN113536015A (en) Cross-modal retrieval method based on depth identification migration
CN110362828B (en) Network information risk identification method and system
CN113407778A (en) Label identification method and device
CN116955591A (en) Recommendation language generation method, related device and medium for content recommendation
CN116977701A (en) Video classification model training method, video classification method and device
CN113626553B (en) Cascade binary Chinese entity relation extraction method based on pre-training model
CN115292533A (en) Cross-modal pedestrian retrieval method driven by visual positioning
Hammad et al. Characterizing the impact of using features extracted from pre-trained models on the quality of video captioning sequence-to-sequence models
Jeevitha et al. Natural language description for videos using NetVLAD and attentional LSTM
CN117789099B (en) Video feature extraction method and device, storage medium and electronic equipment

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WW01 Invention patent application withdrawn after publication

Application publication date: 20220531

WW01 Invention patent application withdrawn after publication