Safer Classification by Synthesis
This addresses safety concerns in classification for AI systems, though it is incremental as it builds on existing generative models like GANs and VAEs.
The paper tackles the problem of deep neural networks being easily fooled by out-of-distribution examples with high confidence, proposing a generative classification method that learns a model per class and queries for similarity at test time, resulting in resilience to such examples while maintaining competitive performance.
The discriminative approach to classification using deep neural networks has become the de-facto standard in various fields. Complementing recent reservations about safety against adversarial examples, we show that conventional discriminative methods can easily be fooled to provide incorrect labels with very high confidence to out of distribution examples. We posit that a generative approach is the natural remedy for this problem, and propose a method for classification using generative models. At training time, we learn a generative model for each class, while at test time, given an example to classify, we query each generator for its most similar generation, and select the class corresponding to the most similar one. Our approach is general and can be used with expressive models such as GANs and VAEs. At test time, our method accurately "knows when it does not know," and provides resilience to out of distribution examples while maintaining competitive performance for standard examples.