Resonator network papers in print - Redwood Center for Theoretical Neuroscience
An exciting new model of factorization in neural circuits, called resonator networks, has been published in the December issue of Neural Computation. Appearing as a two-part series, the articles by Paxon Frady, Spencer Kent, and collaborators, detail how factorization problems arise in cognition and how resonator networks may be a powerful new tool for solving such problems efficiently. They explain how resonator networks define a nonlinear dynamical system with fascinating properties and how using this system to solve an important vector factorization problem is superior to a number of alternative approaches. This opens the door for applications of Vector Symbolic Architectures (VSAs, otherwise known as HD Computing) to a much wider range of problems than previously considered, and is a cornerstone of our ongoing efforts to explore VSAs as framework for cognitive representation in the brain. You can view both articles by visiting our publications page.
Please join us for a seminar, a journal club meeting, or reach out to one of our members directly! Prospective PhD students should apply to the graduate programs of the Helen Wills Neuroscience Institute, Vision Science Graduate Group, or other relevant UC Berkeley departments such as Physics, Electrical Engineering and Computer Sciences, Bioengineering, Biophysics, Statistics or Mathematics.
