Timezone: »

Deep Gamblers: Learning to Abstain with Portfolio Theory
Liu Ziyin · Zhikang Wang · Paul Pu Liang · Russ Salakhutdinov · Louis-Philippe Morency · Masahito Ueda

Wed Dec 11 10:45 AM -- 12:45 PM (PST) @ East Exhibition Hall B + C #43
We deal with the selective classification problem (supervised-learning problem with a rejection option), where we want to achieve the best performance at a certain level of coverage of the data. We transform the original $m$-class classification problem to (m+1)-class where the (m+1)-th class represents the model abstaining from making a prediction due to disconfidence. Inspired by portfolio theory, we propose a loss function for the selective classification problem based on the doubling rate of gambling. Minimizing this loss function corresponds naturally to maximizing the return of a horse race, where a player aims to balance between betting on an outcome (making a prediction) when confident and reserving one's winnings (abstaining) when not confident. This loss function allows us to train neural networks and characterize the disconfidence of prediction in an end-to-end fashion. In comparison with previous methods, our method requires almost no modification to the model inference algorithm or model architecture. Experiments show that our method can identify uncertainty in data points, and achieves strong results on SVHN and CIFAR10 at various coverages of the data.

Author Information

Liu Ziyin (University of Tokyo)
Zhikang Wang (University of Tokyo)
Paul Pu Liang (Carnegie Mellon University)
Russ Salakhutdinov (Carnegie Mellon University)
LP Morency (Carnegie Mellon University)
Masahito Ueda (University of Tokyo)

More from the Same Authors