Brain-Based Machine Learning Algorithms - Alexander Gaura

Abstract

Traditional neural network models are based on ideas from neuroscience that have since become outdated. This paper investigates the performance of new neural network models that are based on modern neuroscience and compares their performance to other models on similar datasets. In particular, these models are useful for unsupervised learning, and differ most in how they receive input, their layering, and their learning rule which is based on Hebbian plasticity.