Deep learning training data augmentation method and device for generating adversarial sample in real time, electronic equipment and medium

Abstract

The invention provides a deep learning training data augmentation method and device for generating an adversarial sample in real time, electronic equipment and a medium, and the method comprises the steps: modifying an input picture through network gradient return, generating the adversarial sample, and training a deep learning network through the generated adversarial sample in real time; using two optimizers, optimizing parameters of an adversarial sample training network and parameters of a deep learning network in training, so that the effect that adversarial parameters and network parameters can be optimized at the same time in one-time loop iteration is achieved, adversarial and training are accelerated, and only one adversarial parameter layer,one loss function and one optimizer needs to be added on the basis of an original network structure ; the robustness of the trained deep learning model is effectively improved, the precision and recall of the model in practical application are improved, and the uninterpretable misjudgment phenomenon of the model on unknown data can be effectively avoided.

Description

technical field [0001] The present invention relates to the technical field of deep learning training methods, and more specifically, to a deep learning training data augmentation, device, electronic equipment and medium for real-time generation of confrontation samples. Background technique [0002] In deep learning model training, because it is not easy to obtain labeled training data, the common practice is to scale, rotate, crop, add noise, projective transformation, superposition, etc. to the training data, and enrich it into the training set. These common operations to enrich the training set are called data augmentation. Data augmentation is to use limited training data to allow the model to learn as much as possible the visual features of the objects in the picture that are independent of position, angle, and noise. [0003] Although the above-mentioned common data augmentation methods can generate a large amount of training data using limited labeled images, they h...

AI Technical Summary

Problems solved by technology

This is also one of the reasons for the over-fitting phenomenon. The so-called over-fitting phenomenon means that the model can achieve high training accuracy on the training set, but the accuracy cannot meet the requirements in actual use.

For example, if a model only uses Gaussian noise and salt-and-pepper noise as data augmentation methods in training, when it encounters image distortion caused by video encoding and decoding in the application, it will have a serious decrease in accuracy.

[0005] Second, due to the black-box characteristics of the deep learning model, the model trainer cannot fully understand the features learned by the model

But its disadvantage is that the adversarial samples generated by this method are aimed at an already trained network, and its algorithm complexity is high, it cannot be combined with the model training process, and it cannot be dynamically adjusted according to the changing network during training. Parameters, constantly targeted to generate adversarial samples, so the scope of application is limited

Images