Skip to main navigationSkip to main content
The University of Southampton
CORMSIS Centre for Operational Research, Management Sciences and Information Systems

Regularization in directable environments with application to Tetris -- Talk by J Malte Lichtenberg (University of Bath) Event

16:00 - 18:00
3 October 2019
Room 3043, Building 2, Southampton Business School, University of Southampton, Highfield Campus, SO17 1BJ

For more information regarding this event, please email Konstantinos Katsikopoulos at .

Event details

Learning statistical models from small data sets is difficult in the absence of specific domain knowledge. We present a regularized linear regression model called shrinking toward equal weights (or STEW), which benefits from a generic and prevalent form of prior knowledge: feature directions. By adaptively shrinking weights toward each other, STEW tries to find the happy medium between the ordinary least squares solution and an equal-weighting model. We provide theoretical results on the equal-weighting solution that explain how STEW can productively trade-off bias and variance, and how knowledge about feature directions helps to reduce its bias even further. Across a wide range of learning problems, including Tetris, STEW outperformed existing regularized linear models, including ridge regression and the Lasso, when feature directions were known. The model proved to be robust to unreliable (or absent) feature directions, outperforming alternative models under diverse conditions. Our results in Tetris were obtained by using a novel approach to learning in sequential decision environments based on multinomial logistic regression.

Speaker information

J Malte Lichtenberg , University of Bath, is a Marie Curie Fellow in the Department of Computer Science and the Centre for Digital Entertainment. Previously, he was working as a statistical programmer at the Center for Adaptive Behaviour and Cognition of the Max Planck Institute for Human Development. He holds a diploma in statistics and economics from ENSAE Paris, and a MSc in statistics from the Humboldt-University of Berlin. Malte’s research interests centre on agents that learn what to do in environments that are characterized by limited resources and high uncertainty. He draws inspiration from models of human decision making.

Privacy Settings