A Less Biased Evaluation of Out-of-distribution Sample Detectors
This addresses the need for reliable evaluation in out-of-distribution detection to prevent unpredictable behavior in deployed learning systems, though it is incremental as it focuses on improving evaluation rather than proposing a new detection method.
The paper tackled the problem of evaluating out-of-distribution sample detectors by introducing OD-test, a three-dataset evaluation scheme, and found that existing techniques have low accuracy and are unreliable for realistic high-dimensional image applications.
In the real world, a learning system could receive an input that is unlike anything it has seen during training. Unfortunately, out-of-distribution samples can lead to unpredictable behaviour. We need to know whether any given input belongs to the population distribution of the training/evaluation data to prevent unpredictable behaviour in deployed systems. A recent surge of interest in this problem has led to the development of sophisticated techniques in the deep learning literature. However, due to the absence of a standard problem definition or an exhaustive evaluation, it is not evident if we can rely on these methods. What makes this problem different from a typical supervised learning setting is that the distribution of outliers used in training may not be the same as the distribution of outliers encountered in the application. Classical approaches that learn inliers vs. outliers with only two datasets can yield optimistic results. We introduce OD-test, a three-dataset evaluation scheme as a more reliable strategy to assess progress on this problem. We present an exhaustive evaluation of a broad set of methods from related areas on image classification tasks. Contrary to the existing results, we show that for realistic applications of high-dimensional images the previous techniques have low accuracy and are not reliable in practice.