SEAL: Searching Expandable Architectures for Incremental Learning

TL;DR

SEAL introduces a NAS-based framework for incremental learning that adaptively expands neural architectures only when necessary, balancing plasticity and stability while reducing model size and forgetting.

Contribution

It presents a novel NAS-based method that dynamically adjusts model size based on capacity estimation, improving efficiency and performance in data-incremental learning.

Findings

Reduces forgetting and improves accuracy across benchmarks.

Maintains smaller model sizes compared to existing methods.

Effectively adapts architecture only when needed.

Abstract

Incremental learning is a machine learning paradigm where a model learns from a sequential stream of tasks. This setting poses a key challenge: balancing plasticity (learning new tasks) and stability (preserving past knowledge). Neural Architecture Search (NAS), a branch of AutoML, automates the design of the architecture of Deep Neural Networks and has shown success in static settings. However, existing NAS-based approaches to incremental learning often rely on expanding the model at every task, making them impractical in resource-constrained environments. In this work, we introduce SEAL, a NAS-based framework tailored for data-incremental learning, a scenario where disjoint data samples arrive sequentially and are not stored for future access. SEAL adapts the model structure dynamically by expanding it only when necessary, based on a capacity estimation metric. Stability is preserved through cross-distillation training after each expansion step. The NAS component jointly searches for both the architecture and the optimal expansion policy. Experiments across multiple benchmarks demonstrate that SEAL effectively reduces forgetting and enhances accuracy while maintaining a lower model size compared to prior methods. These results highlight the promise of combining NAS and selective expansion for efficient, adaptive learning in incremental scenarios.