Identification model training method and device, and information identification method and device

Abstract

The invention discloses an identification model training method and device and an information identification method and device. The method comprises the steps that a to-be-processed target sample set is acquired, and the target sample set comprises a plurality of initial data samples corresponding to a target problem type; determining similar data samples of the initial data samples from the target sample set; fusing the initial data sample and the similar data sample to generate a target data sample; and training a preset neural network model by using the target data sample and the initial data sample to obtain an identification model. According to the embodiment of the invention, the similar data sample corresponding to each initial data sample in the sample set is determined, and then the initial data samples and the similar data samples are fused to obtain the target data sample. The number of samples in the sample set can be increased through the target data samples, training balance in the model training process is guaranteed, and the problem that in the prior art, the number of center samples is unbalanced, and consequently the accuracy is low after model training is solved.

Description

technical field [0001] The present application relates to the field of artificial intelligence, in particular to a training method and device for a recognition model, and an information recognition method and device. Background technique [0002] Intent recognition is essentially a classification problem, which can be achieved by using rule-based, traditional machine learning algorithms, and deep learning algorithms. For intent recognition tasks with a small amount of training data, the BERT model performs well. The BERT model refers to a transformer-based bidirectional encoder representation technology, which can be trained on large-scale natural language data and on the basis of the training model. The model is fine-tuned for specific training tasks to achieve intent recognition tasks, so the BERT model is relatively friendly to small data set model training. During the training process, the sample size of a certain problem type is greater than the sum of the correspondin...

AI Technical Summary

Problems solved by technology

During the training process, the sample size of a certain problem type is larger than the sum of the corresponding sample sizes of the other problem types, resulting in a serious imbalance in the sample size used for training, which will cause the final recognition model to tilt towards the problem type with a large sample size.

Images