A Biologically Realistic Approach Using Pulse-Based Computational Architectures To Tackle The Visual Categorization Challenge

Vision inference is the process of the eye, recognizing an object, analyzing the same and reconstructing a complete picture of an object. Unlike some mental processes like computing an addition, or reasoning for a decision in a chess game, vision inference is a mental ability that the person involved is not consciously aware of. As a result, many people including experienced researchers often underestimate the difficulty of recognizing the image while modeling the same. The earliest pattern recognition research dates back to 1966 when Marvin Minsky and Seymour Papert tried developing an object recognition system (Papert 1966). Yet it has taken more than 50 years to develop a system for image classification which is similar in accuracy to an average person. In the last decade, machine learning has succeeded in tasks like identifying a wide variety of objects and faces in realistic situations and contexts. The key to success is the focus of shift to object identification, based on the developments observed in the field of neuroscience.