Timezone: »

 
Poster
An Unbiased Risk Estimator for Learning with Augmented Classes
Yu-Jie Zhang · Peng Zhao · Lanjihong Ma · Zhi-Hua Zhou

Wed Dec 09 09:00 PM -- 11:00 PM (PST) @ Poster Session 4 #1213
in Poster Session 5 »

This paper studies the problem of learning with augmented classes (LAC), where augmented classes unobserved in the training data might emerge in the testing phase. Previous studies generally attempt to discover augmented classes by exploiting geometric properties, achieving inspiring empirical performance yet lacking theoretical understandings particularly on the generalization ability. In this paper we show that, by using unlabeled training data to approximate the potential distribution of augmented classes, an unbiased risk estimator of the testing distribution can be established for the LAC problem under mild assumptions, which paves a way to develop a sound approach with theoretical guarantees. Moreover, the proposed approach can adapt to complex changing environments where augmented classes may appear and the prior of known classes may change simultaneously. Extensive experiments confirm the effectiveness of our proposed approach.

Author Information

Yu-Jie Zhang (Nanjing University)
Peng Zhao (Nanjing University)
Lanjihong Ma (Nanjing University)
Zhi-Hua Zhou (Nanjing University)

More from the Same Authors