Student Research Presentations
Date: 3:30pm - 4:30pm PDT September 27, 2011 Location: JR Howard 244
JR Howard 244
A Linear Classifier Outperforms UCT in 9x9 Go
Presented by Sam Dodson (’13)
The dominant paradigm in computer Go is Monte-Carlo Tree Search (MCTS). This technique chooses a move by playing a series of simulated games, building a search tree along the way. After many simulated games, the most promising move is played. We propose replacing the search tree with a neural network. Where previous neural network Go research has used the state of the board as input, our network uses the last two moves. In experiments exploring the effects of various parameters, our network outperforms a generic MCTS player that uses the Upper Confidence bounds applied to Trees (UCT) algorithm. A simple linear classifier performs even better.
Leveraging Pathway Knowledge for Cancer Treatment
Presented by Christopher Anderson (’13)
Our project used publicly available protein interactions and canonical pathways to construct a human interactome for insight to important prostate cancer signaling events. In order to better understand the shape and structure of our interactome, we used graphical visualization programs to display data in both 2D and 3D perspectives. Next, we began mapping siRNA knockdown data from cancer cell cultures to our interactome to look for pathways with more connected knockdown hits than would be expected by chance, indicating their potential for therapeutic study. Results illustrate the “Aurora B signaling” pathway, containing 35 edges connecting siRNA knockdown hits between the 12 knockdown hits, was significantly enriched (p-value <= .0001) and warrants further investigation.