2019/02/16 by Marc Maliar, Maliar, Marc
Computer Science · Neuroscience · #Artificial Intelligence (cs.AI) #Cognitive Science and Education Research #Computational Physics and Python Applications #Explainable Artificial Intelligence (XAI) #FOS: Computer and information sciences #Neural Networks and Applications
paper · pdf · doi:10.48550/arxiv.1903.03408
openalex publication_date 2019/02/16 · openalex created_date 2022/07/29 · openalex updated_date 2026/07/28
I use simulation of two multilayer neural networks to gain intuition into the\ndeterminants of human learning. The first network, the teacher, is trained to\nachieve a high accuracy in handwritten digit recognition. The second network,\nthe student, learns to reproduce the output of the first network. I show that\nlearning from the teacher is more effective than learning from the data under\nthe appropriate degree of regularization. Regularization allows the teacher to\ndistinguish the trends and to deliver "big ideas" to the student. I also model\nother learning situations such as expert and novice teachers, high- and\nlow-ability students and biased learning experience due to, e.g., poverty and\ntrauma. The results from computer simulation accord remarkably well with\nfinding of the modern psychological literature. The code is written in MATLAB\nand will be publicly available from the author's web page.\n