Current Lab Members
Biocomputing Lab members for spring 2017: (left to right) Jewel Chu, Amy Rapin, Michael Vaschillo, Michael Stiber, Fumitaka Kawasaki, Destiny Boyer, Tom Wong.
Michael Stiber, Laboratory Director, Professor of Computing & Software Systems and Associate Dean for Research and Graduate Studies, UW Bothell School of STEM. Prof. Stiber has over 20 years' experience in Biocomputing, including extensive collaborations with neuroscientists and biomathematicians. He has developed software to support neural simulation and data analysis for a wide range of platforms.
Shelby Eaton is a Software Engineering Consultant at Common Sense Systems, Inc., a software engineering services firm in Woodinville, Washington. A 2000 graduate of the CSS program, Ms. Eaton holds Bachelor of Science degrees in Computing & Software Systems and in Mathematics. She has recently worked on software for enterprise information portal customizations and martial arts scoring systems. She enjoys studying the connections among mathematics, computer science, and music, especially as it relates to neural networks.
Haruka Higuchi, Hanjoo Kim, Hyon Kim, and Kathleen Pollmann developed a prototype of a "signal computing" software laboratory for use in undergraduate laboratory classes.
Thomas Holderman simulated neurons under varying degrees of input accuracy and error rates to see if neurons can utilize error correcting codes.
Mark Molina worked on a program to create 3-D models of a neuron from stacks of 2-D photos.
Mark Pottorf built a simulation to capture state data from a neuron presented with low-precision inputs.
Patricia Tressel is a Ph.D. student in the UW Seattle Department of Computer Science and Engineering. She describes her own work as: "Neurons are notoriously 'noisy'. Their responses depend on the signals they've received and how they're 'feeling' at the moment (their internal state), but characteristics of their response signals (e.g. strength and timing) are describable by probability distributions. The usual methods for simulating the operation of interconnected neurons use definite times and strengths for signals, and either ignore the apparent unreliability, or run the simulation over and over making different random choices of signal characteristics on each simulation run. Instead, we'd like to carry entire probability distributions analytically through the simulation. Then we'd only have to run the simulation once, and we'd have the probability distribution of the results. This will provide much faster simulations. Additionally, knowing the response distribution of a given neural circuit tells us whether it can constrain the randomness in its response sufficiently that the response can still be meaningful."