Sponsor
This material is based upon work supported by the National Science Foundation under grant no. 1028120.
Document Type
Pre-Print
Publication Date
4-3-2014
Subjects
Computational neuroscience, Pattern recognition systems, Perceptrons -- Analysis, Biochemical phenomena -- Experiments, Biomathematics, Neural networks (Computer science)
Abstract
The current biochemical information processing systems behave in a predetermined manner because all features are defined during the design phase. To make such unconventional computing systems reusable and programmable for biomedical applications, adaptation, learning, and self-modification baaed on external stimuli would be highly desirable. However, so far, it haa been too challenging to implement these in real or simulated chemistries. In this paper we extend the chemical perceptron, a model previously proposed by the authors, to function as an analog instead of a binary system. The new analog asymmetric signal perceptron learns through feedback and supports MichaelisMenten kinetics. The results show that our perceptron is able to learn linear and nonlinear (quadratic) functions of two inputs. To the best of our knowledge, it is the first simulated chemical system capable of doing so. The small number of species and reactions allows for a mapping to an actual wet implementation using DNA-strand displacement or deoxyribozymes. Our results are an important step toward actual biochemical systems that can learn and adapt.
Persistent Identifier
http://archives.pdx.edu/ds/psu/12176
Citation Details
Banda, Peter, and Christof Teuscher. "Learning Two-input Linear and Nonlinear Analog Functions with a Simple Chemical System." arXiv preprint arXiv:1404.0427 (2014)
Description
This is the author's version of a paper to be presented at the Unconventional Computation & Natural Computation conference, to be held July 14-18 at the University of Western Ontario, London, Ontario, Canada.