Transactions on Machine Learning Research 08/2022
How do we know when the predictions made by a classifier can be trusted? This
is a fundamental problem that also has immense practical applicability,
especially in safety-critical areas such as medicine and autonomous driving.
The de facto approach of using the classifier's softmax outputs as a proxy for
trustworthiness suffers from the over-confidence issue; while the most recent
works incur problems such as additional retraining cost and accuracy versus
trustworthiness trade-off. In this work, we argue that the trustworthiness of a
classifier's prediction for a sample is highly associated with two factors: the
sample's neighborhood information and the classifier's output. To combine the
best of both worlds, we design a model-agnostic post-hoc approach NeighborAgg
to leverage the two essential information via an adaptive neighborhood
aggregation. Theoretically, we show that NeighborAgg is a generalized version
of a one-hop graph convolutional network, inheriting the powerful modeling
ability to capture the varying similarity between samples within each class. We
also extend our approach to the closely related task of mislabel detection and
provide a theoretical coverage guarantee to bound the false negative.
Empirically, extensive experiments on image and tabular benchmarks verify our
theory and suggest that NeighborAgg outperforms other methods, achieving
state-of-the-art trustworthiness performance.
Miao Xiong, Shen Li, Wenjie Feng, Ailin Deng, Jihai Zhang, Bryan Hooi
2022-11-29
