OpenIncrement: A Unified Framework for Open Set Recognition and Deep Class-Incremental Learning

TL;DR

OpenIncrement presents a unified framework combining deep class-incremental learning with open set recognition, improving the detection of novel samples and enhancing model performance in dynamic environments.

Contribution

It introduces a novel integrated approach that refines features for both incremental learning and open set recognition, addressing misclassification issues in practical scenarios.

Findings

Outperforms existing incremental learning methods

Achieves superior open set recognition accuracy

Enhances detection of novel samples in dynamic settings

Abstract

In most works on deep incremental learning research, it is assumed that novel samples are pre-identified for neural network retraining. However, practical deep classifiers often misidentify these samples, leading to erroneous predictions. Such misclassifications can degrade model performance. Techniques like open set recognition offer a means to detect these novel samples, representing a significant area in the machine learning domain. In this paper, we introduce a deep class-incremental learning framework integrated with open set recognition. Our approach refines class-incrementally learned features to adapt them for distance-based open set recognition. Experimental results validate that our method outperforms state-of-the-art incremental learning techniques and exhibits superior performance in open set recognition compared to baseline methods.