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

arXiv: https://arxiv.org/abs/2606.09960 · HTML: https://mlsanity.com/paper/2606.09960/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 12 of its 18 references that are in the map.
- First posted on arXiv 8 Jun 2026.

## Builds on (most cited first, 12 of 12)

- [DER: Dynamically Expandable Representation for Class Incremental Learning](https://mlsanity.com/paper/2103.16788/) (2021)
- [CORe50: a New Dataset and Benchmark for Continuous Object Recognition](https://mlsanity.com/paper/1705.03550/) (2017)
- [FOSTER: Feature Boosting and Compression for Class-Incremental Learning](https://mlsanity.com/paper/2204.04662/) (2022)
- [Class-Incremental Learning: A Survey](https://mlsanity.com/paper/2302.03648/) (2023)
- [REMIND Your Neural Network to Prevent Catastrophic Forgetting](https://mlsanity.com/paper/1910.02509/) (2019)
- [A Model or 603 Exemplars: Towards Memory-Efficient Class-Incremental Learning](https://mlsanity.com/paper/2205.13218/) (2022)
- [FeTrIL: Feature Translation for Exemplar-Free Class-Incremental Learning](https://mlsanity.com/paper/2211.13131/) (2022)
- [Self-Sustaining Representation Expansion for Non-Exemplar Class-Incremental Learning](https://mlsanity.com/paper/2203.06359/) (2022)
- [RMM: Reinforced Memory Management for Class-Incremental Learning](https://mlsanity.com/paper/2301.05792/) (2023)
- [How Efficient Are Today's Continual Learning Algorithms?](https://mlsanity.com/paper/2303.18171/) (2023)
- [Vision Transformers on the Edge: A Comprehensive Survey of Model Compression and Acceleration Strategies](https://mlsanity.com/paper/2503.02891/) (2025)
- [CIFNet: An Analytic Neural Learning Framework for Efficient and Calibrated Class-Incremental Learning](https://mlsanity.com/paper/2509.11285/) (2025)

## Abstract (arXiv)

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.