Linguistic instructions learning based on associative memories and its application to a facial model

Human learning on linguistic level is superior to other kinds of learning. If a neural network can be trained by natural language instead of numerical data, we can train machines as well as human beings without detailed training data. Conventional neural networks need numerical data to be trained. On the other hand, a linguistic level learning is able to train machines as if they were human beings. In this paper, we propose a linguistic instructions learning method based on a fuzzy associative memory network, which acquires knowledge from natural language, and we refine a facial expressions model by means of this method.