Few-shot Image Sentiment Classification Method Based on Meta-learning

Abstract

The invention discloses a meta-learning-based emotion classification method for few-sample images. Firstly, a plurality of meta-learning tasks similar to the few-sample images with emotional label information in the target data set are constructed on the source data set, and then these meta-learning tasks are Task learning, to obtain a good classification model initialization parameters, so that the classification model can achieve better classification results when facing emotional images in a few-sample target data set. The present invention can not only greatly alleviate the need for labeling data, but also the method based on meta-learning is more in line with the human learning method (human learning new tasks are all based on the tasks that have been learned), and can make the neural network model more intelligent .

Description

technical field [0001] The invention belongs to the field of neural networks (Neural Networks, NN) in machine learning (Machine Learning), relates to an image emotion classification method based on deep learning, in particular to a few-sample image emotion classification based on meta-learning. Background technique [0002] Some psychological studies have shown that human emotional responses vary with different visual stimuli, and in particular, pictures are very important for attracting people's attention and motivating them to take action. Various previous studies have shown that, for some consumers, pictures can lead to a perception of higher product quality without being able to actually touch the product, making this visual cue all the more important. In e-commerce scenarios, images also have an impact on buyer intent, trust, risk reduction, conversion and click-through rates. In some charitable fundraisers, the combination of positivity and negativity in the "donation...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the problem that the current deep learning method for image emotion is difficult to achieve learning based on limited resources, the purpose of the present invention is to provide an image emotion classification method based on meta-learning, which can realize the emotion classification of few-sample images

Images