OVOR: OnePrompt with Virtual Outlier Regularization for Rehearsal-Free Class-Incremental Learning

2024-02-06

International Conference on Learning Representations

AI Summary

In rehearsal-free class-incremental learning (CIL), severe inter-task category confusion arises due to ambiguous decision boundaries, while existing prompt-based methods rely on task-specific prompt pools, incurring substantial computational overhead. Method: We propose Virtual Outlier Penalization Regularization (VOPR) and OnePrompt. VOPR synthesizes semantically consistent virtual outlier samples to explicitly tighten classifier decision boundaries and mitigate boundary-induced confusion; OnePrompt employs a single shared learnable prompt—eliminating the need for multiple task-specific prompt pools. Contribution/Results: We provide the first theoretical analysis and empirical validation demonstrating that a single prompt can match state-of-the-art multi-prompt methods in rehearsal-free CIL. On ImageNet-R and CIFAR-100, our approach significantly outperforms mainstream rehearsal-free CIL baselines, reducing parameter count by ~90% and inference latency by ~40%, thereby achieving a unified trade-off between high accuracy and high efficiency.

Technology Category

• Natural Language Processing: Prompt Engineering / Prompting
• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Recent works have shown that by using large pre-trained models along with learnable prompts, rehearsal-free methods for class-incremental learning (CIL) settings can achieve superior performance to prominent rehearsal-based ones. Rehearsal-free CIL methods struggle with distinguishing classes from different tasks, as those are not trained together. In this work we propose a regularization method based on virtual outliers to tighten decision boundaries of the classifier, such that confusion of classes among different tasks is mitigated. Recent prompt-based methods often require a pool of task-specific prompts, in order to prevent overwriting knowledge of previous tasks with that of the new task, leading to extra computation in querying and composing an appropriate prompt from the pool. This additional cost can be eliminated, without sacrificing accuracy, as we reveal in the paper. We illustrate that a simplified prompt-based method can achieve results comparable to previous state-of-the-art (SOTA) methods equipped with a prompt pool, using much less learnable parameters and lower inference cost. Our regularization method has demonstrated its compatibility with different prompt-based methods, boosting those previous SOTA rehearsal-free CIL methods' accuracy on the ImageNet-R and CIFAR-100 benchmarks. Our source code is available at https://github.com/jpmorganchase/ovor.

Problem

• Mitigates confusion between classes from different incremental tasks
• Eliminates need for prompt pools to reduce computation costs
• Improves rehearsal-free class-incremental learning accuracy on benchmarks

Innovation

• Virtual outlier regularization tightens decision boundaries
• Simplified prompt-based method reduces parameters and cost
• Compatible with various prompt-based methods, boosting accuracy