Manuscript Summary
Classification of image, text, and audio is one of the basic tasks of supervised machine learning. In recent years, deep neural networks have become the defacto approach to solving such supervised problems. Deep learning allows for the automated learning of features. In this work, we explore an older method known as cellular automata as an alternative approach to learning and generating features used for classification. Specifically, we use the concept of dynamical phase transitions to explore a fundamentally different approach to classification.
Open Access Link
