Wolfram Computation Meets Knowledge

Wolfram Summer School


Mi Zhang

Science and Technology

Class of 2016


I was previously studying the egg-laying circuit in C. elegans, and later became a programmer.

Project: Using Symbolic and Simple Computation on Handwritten Digits Recognition

My project is to develop a symbolic/conceptual computing neural network mainly consisting of logical gates to recognize the handwritten numeric digits 0–9. That type of neural network is in contrast with the currently widely used analog neural networks in AI. The purpose of this project is to show, if successful, that logical gates composed of neural networks can accomplish interesting tasks. That hopefully will invoke more interest in applying logical gates in AI.

The reason I am interested in logical gates composed of neural networks and symbolic/conceptual computing is because of my previous experimental results of a biological neural network, the egg-laying circuit in C. elegans. Even though that is a very simple circuit, it turns out that all three neurons that control the egg-laying behavior function as a logical gate, and each node encodes a meaningful “symbol” or “concept.” My interest is in seeing whether this type of computational model selected by nature has any advantage and application in AI.