HydraCIL: Decoupled Class-Incremental Learning through Prototype-Guided Multi-Head Classifiers

2026-06-08

AI Summary

This work addresses the challenges of class-incremental learning in resource-constrained settings—namely, frequent retraining, high computational costs, and deployment difficulties—by proposing a decoupled continual learning framework. The approach freezes the backbone network and trains only lightweight, task-specific classification heads, augmented with a prototype-guided dynamic task selection mechanism. By eliminating the need to repeatedly retrain the feature extractor, the method substantially reduces training time and energy consumption. Evaluated on standard benchmarks such as CIFAR-100 and ImageNet-100, it achieves performance on par with or superior to state-of-the-art methods while significantly lowering its carbon footprint, thereby balancing efficiency and accuracy.

Abstract

We present HydraCIL, a decoupled continual learning model based on prototype-guided multi-head classifiers, targeting sustainable deployment in embedded and resource-constrained environments. While most Class-Incremental Learning (CIL) methods rely on powerful hardware and long retraining cycles, real-world systems, such as robots or edge AI devices, must adapt quickly with limited resources. HydraCIL addresses this gap by freezing the backbone and decoupling feature extraction from learning. For each task, features are extracted once and a lightweight, task-specific classifier head is created, avoiding costly backbone retraining. At inference, HydraCIL selects the appropriate head via similarity with prototypes. Experiments on CIFAR-100, ImageNet-100, CoRe50, and Flowers102 datasets show that HydraCIL matches or outperforms state-of-the-art CIL methods while significantly reducing training time and carbon footprint, making it a practical solution for continual learning in real-world and embedded settings, where energy efficiency and rapid adaptation are critical.

Problem

• Class-Incremental Learning
• resource-constrained environments
• continual learning
• embedded systems
• energy efficiency

Innovation

• Class-Incremental Learning
• Decoupled Learning
• Prototype-Guided Classification
• Multi-Head Classifiers
• Energy-Efficient AI