Class-incremental Learning via Deep Model Consolidation

TL;DR

This paper introduces Deep Model Consolidation, a method for class-incremental learning that effectively combines models trained on old and new classes using unlabeled auxiliary data, avoiding catastrophic forgetting.

Contribution

It proposes a novel double distillation training approach that consolidates separate models without needing original training data, improving incremental learning performance.

Findings

Outperforms state-of-the-art methods on CIFAR-100 and CUB-200 image classification tasks.

Achieves superior results in object detection on PASCAL VOC 2007.

Effective even when original training data is unavailable.

Abstract

Deep neural networks (DNNs) often suffer from "catastrophic forgetting" during incremental learning (IL) --- an abrupt degradation of performance on the original set of classes when the training objective is adapted to a newly added set of classes. Existing IL approaches tend to produce a model that is biased towards either the old classes or new classes, unless with the help of exemplars of the old data. To address this issue, we propose a class-incremental learning paradigm called Deep Model Consolidation (DMC), which works well even when the original training data is not available. The idea is to first train a separate model only for the new classes, and then combine the two individual models trained on data of two distinct set of classes (old classes and new classes) via a novel double distillation training objective. The two existing models are consolidated by exploiting publicly available unlabeled auxiliary data. This overcomes the potential difficulties due to the unavailability of original training data. Compared to the state-of-the-art techniques, DMC demonstrates significantly better performance in image classification (CIFAR-100 and CUB-200) and object detection (PASCAL VOC 2007) in the single-headed IL setting.