Liang  Zhou

Liang Zhou

Scholar Title

MIT EECS | MITRE Undergraduate Research and Innovation Scholar

Research Title

An Intuitive Constraint-Based Model of Physical Dynamics

Cohort

2017–2018

Department

EECS

Research Areas
  • Artificial Intelligence & Machine Learning
Supervisor

Una-May O'Reilly

Abstract

How do we as humans develop intuition about the physical world around us? It is obvious that letting go of an apple causes it to fall, but it is harder to predict the trajectories that result from two billiard balls colliding. Regardless, we seem to have developed some mental intuition for what will likely happen. Our project seeks to model this cognitive process from an evolutionary computing perspective by introducing genetic programming methods to the development and learning of transferrable physical intuition. By searching over a space of possible programs, the model should be able to generate its own simple forms and respond to novel situations subject to noise and various heuristics. We seek to reinforce the idea that humans possess an abstract framework for physical learning.

Quote

I’ll be developing cognitive models of physical concept learning as part of my research in SuperUROP. I’ve been researching computational theories of mind for more than a year now, and I’m glad to be able to continue working in such an interdisciplinary field and advancing our knowledge about human thought.

Back to Scholars