CONCLAD: COntinuous Novel CLAss Detector

TL;DR

CONCLAD is a continual learning method that detects and learns new classes in post-deployment data using uncertainty estimation, pseudo-labeling, and minimal supervision, improving ongoing class separation.

Contribution

It introduces a novel continual novel class detection approach that operates without oracles, utilizing iterative uncertainty estimation and minimal supervision for effective class discovery.

Findings

Effective separation of novel and old classes across datasets

Robustness demonstrated through ablation studies

Continuous learning with minimal supervision

Abstract

In the field of continual learning, relying on so-called oracles for novelty detection is commonplace albeit unrealistic. This paper introduces CONCLAD ("COntinuous Novel CLAss Detector"), a comprehensive solution to the under-explored problem of continual novel class detection in post-deployment data. At each new task, our approach employs an iterative uncertainty estimation algorithm to differentiate between known and novel class(es) samples, and to further discriminate between the different novel classes themselves. Samples predicted to be from a novel class with high-confidence are automatically pseudo-labeled and used to update our model. Simultaneously, a tiny supervision budget is used to iteratively query ambiguous novel class predictions, which are also used during update. Evaluation across multiple datasets, ablations and experimental settings demonstrate our method's effectiveness at separating novel and old class samples continuously. We will release our code upon acceptance.