University of Wisconsin - Green Bay


Research Council

Research
Success

Peter Breznay

Peter Breznay

Fall 2010

"Modeling and Simulation of Embedded Circuit Components in Robotic Intelligence"

Grant in Aid of Research

Final Report: "We investigate the proposition that every genuine scientific problem is, ultimately, a computational problem. Phenomena in the physical, biological and social spheres of existence, can, theoretically, be modeled and simulated in a computing device with arbitrary precision, given unlimited computational resources (which are not available in practice). In addition, every phenomenon can be interpreted as a computation (running of a program), performed by some sub-system of the physical universe. Starting from this computational perspective, we propose that the problem of consciousness can also be investigated as a computational problem, in carrying out a scientific program aimed at building artificial entities that show verifiable emergence of consciousness. Our hypothesis is that consciousness in an artificial device is unlikely to manifest itself originating in a detached computer, but rather in a humanoid robotic device that has full sensory capacities and is capable both of active interaction with its physical environment and of full sentence communication with humans. As a result, we propose that in an attempt to create artificial, man-made devices that are capable of acquiring true consciousness, we need to build a small society of fully sentient robots that are maximally inter-operational with their environment, with humans and with each other, that have the following features: 1. A learning-enabled, self-restructuring, artificial neural network-based central cognitive organ ("brain"). 2. A global communication subsystem of the brain that separates conscious sensation from unconscious by filtering out conflicting sensations, in order to form a coherent view of the robots' environment. 3. The ability of abstraction by generalization to allow memory forming, storage and retrieval. 4. The ability to recognize and synthesize human language, in form of full sentence communication. We present preliminary simulation results regarding the learning ability aspect of the proposed central neural organ. In the simulation we use a mirror neuron mechanism to reproduce rudimentary social learning, performed by a “child” neural network of a “parent” neural network and achieved by a form of imitation-based learning."


Spring 2008

"The Heliotropic Worm: Establishing Theory of Mind in Artificial Neural Networks"

Grant for Integrating Research and Teaching


Spring 2008

"Transport and Replication Dynamics of Diffusion Processes in Networked Systems"

Grant in Aid of Research


Spring 2005

"Hardware Implementation of back-propagating artificial neural network based pattern recognizer"

Grant for Integrating Research and Teaching


Spring 2005

"A Dynamic Network Flow Theoretical Model of Information Diffusion"

Grant in Aid of Research


Spring 2003

"Travel to International Conference on Parallel Processing"

Grant in Aid of Research


Spring 2002

"Neural Network Proposal"

Grant for Integrating Research and Teaching


Fall 2000

"Recursive Heirarchical Interconnection Networks with Asymptomatically Optimal Parameters"

Grant in Aid of Research


Fall 1999

"Optimizing the Indexing of Continuously Moving Objects"

Grant in Aid of Research