Imagining how a cell thinks: The design of reaction network schemes that do machine l earning
A living cell responds in sophisticated ways to its environment. Such behavior is all the more remarkable when one considers that a cell is a bag of molecules. A detailed algorithmic explanation is required for how a network of chemical reactions can produce sophisticated behavior. Though several previous works have shown that reaction networks are computationally universal and can in principle implement any algorithm, there is scope for constructions that map well onto biological reality, make efficient use of the computational potential of the native dynamics of reaction networks, and make contact with statistical mechanics.We find that the mathematical structure of reaction networks is particularly well suited to implementing modern machine learning algorithms. We describe a new reaction network scheme for solving a l arge class of statistical problems including the problem of how a cell would infer its environment from receptor-ligand bindings. Specificially we show how reaction networks can implement information projection, and consequently a generalized Expectation-Maximization algorithm, to solve maximum likelihood estimation problems in partially-observed exponential families on categorical data. Our scheme can be thought of as an algorithmic embodiment of E. T. Jaynes's vision of statistical mechanics as statistical inference.
Manoj Gopalkrishnan is Associate Professor in the Department of Electrical Engineering at IIT Bombay. His Ph. D. work was in Computer Science under Professor Leonard Adleman at University of Southern California. His research interests span theoretical ideas in, and connections between, Machine Learning, Thermodynamics, and the Information Sciences, with applicatio ns to Molecular Computing and Biological Systems. He is unable to resist dabbling in new research areas, and his latest infatuations include Biomechanics, Resource Theories of Quantum Thermodynamics, Neuroscience, and Multi Agent Reinforcement Learning.