T2FNorm: Extremely Simple Scaled Train-time Feature Normalization for OOD Detection
This addresses the challenge of safe deployment of neural networks in real-world applications by enhancing OOD detection, though it appears incremental as it builds on existing feature normalization techniques.
The paper tackled the problem of neural networks being overconfident predictors by introducing T2FNorm, a train-time feature normalization method that transforms features to hyperspherical space, which substantially improves out-of-distribution detection performance without compromising in-distribution accuracy.
Neural networks are notorious for being overconfident predictors, posing a significant challenge to their safe deployment in real-world applications. While feature normalization has garnered considerable attention within the deep learning literature, current train-time regularization methods for Out-of-Distribution(OOD) detection are yet to fully exploit this potential. Indeed, the naive incorporation of feature normalization within neural networks does not guarantee substantial improvement in OOD detection performance. In this work, we introduce T2FNorm, a novel approach to transforming features to hyperspherical space during training, while employing non-transformed space for OOD-scoring purposes. This method yields a surprising enhancement in OOD detection capabilities without compromising model accuracy in in-distribution(ID). Our investigation demonstrates that the proposed technique substantially diminishes the norm of the features of all samples, more so in the case of out-of-distribution samples, thereby addressing the prevalent concern of overconfidence in neural networks. The proposed method also significantly improves various post-hoc OOD detection methods.