MetaMax: Improved Open-Set Deep Neural Networks via Weibull Calibration

TL;DR

MetaMax is a novel post-processing method for open-set recognition in deep neural networks that enhances calibration by directly modeling class activation vectors, outperforming previous methods like OpenMax.

Contribution

MetaMax introduces a more effective calibration technique that eliminates the need for class mean activation vectors, improving open-set recognition performance.

Findings

MetaMax outperforms OpenMax in open-set recognition tasks.

MetaMax achieves comparable results to state-of-the-art methods.

MetaMax simplifies the calibration process by directly modeling class activation vectors.

Abstract

Open-set recognition refers to the problem in which classes that were not seen during training appear at inference time. This requires the ability to identify instances of novel classes while maintaining discriminative capability for closed-set classification. OpenMax was the first deep neural network-based approach to address open-set recognition by calibrating the predictive scores of a standard closed-set classification network. In this paper we present MetaMax, a more effective post-processing technique that improves upon contemporary methods by directly modeling class activation vectors. MetaMax removes the need for computing class mean activation vectors (MAVs) and distances between a query image and a class MAV as required in OpenMax. Experimental results show that MetaMax outperforms OpenMax and is comparable in performance to other state-of-the-art approaches.