Chinese affective analysis aims to dig out the subjective emotion in Chinese text.At present,most Chinese affective analysis models based on deep learning need to rely on large-scale labeled data for training.Meanwhile,deep learning models are easy to be affected by adversarial disturbance in practical applications,resulting in the degradation of model performance.In response to the above issues,this paper proposes a Chinese affective analysis model based on model-agnostic meta-learning and antagonistic training,which can accelerate the convergence of the model using meta learning under small-scale datasets,and generate confrontation samples to conduct confrontat-ion training on the model,improving the anti-interference ability of the model.Experiments have shown that the model has achieved excellent performance.
维普
万方数据
