Autoencoder-Based Hybrid Replay for Class-Incremental Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Nori, Milad Khademi, Kim, Il-Min, Wang, Guanghui
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915289567854592
author Nori, Milad Khademi
Kim, Il-Min
Wang, Guanghui
author_facet Nori, Milad Khademi
Kim, Il-Min
Wang, Guanghui
contents In class-incremental learning (CIL), effective incremental learning strategies are essential to mitigate task confusion and catastrophic forgetting, especially as the number of tasks $t$ increases. Current exemplar replay strategies impose $\mathcal{O}(t)$ memory/compute complexities. We propose an autoencoder-based hybrid replay (AHR) strategy that leverages our new hybrid autoencoder (HAE) to function as a compressor to alleviate the requirement for large memory, achieving $\mathcal{O}(0.1 t)$ at the worst case with the computing complexity of $\mathcal{O}(t)$ while accomplishing state-of-the-art performance. The decoder later recovers the exemplar data stored in the latent space, rather than in raw format. Additionally, HAE is designed for both discriminative and generative modeling, enabling classification and replay capabilities, respectively. HAE adopts the charged particle system energy minimization equations and repulsive force algorithm for the incremental embedding and distribution of new class centroids in its latent space. Our results demonstrate that AHR consistently outperforms recent baselines across multiple benchmarks while operating with the same memory/compute budgets. The source code is included in the supplementary material and will be open-sourced upon publication.
format Preprint
id arxiv_https___arxiv_org_abs_2505_05926
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Autoencoder-Based Hybrid Replay for Class-Incremental Learning
Nori, Milad Khademi
Kim, Il-Min
Wang, Guanghui
Machine Learning
In class-incremental learning (CIL), effective incremental learning strategies are essential to mitigate task confusion and catastrophic forgetting, especially as the number of tasks $t$ increases. Current exemplar replay strategies impose $\mathcal{O}(t)$ memory/compute complexities. We propose an autoencoder-based hybrid replay (AHR) strategy that leverages our new hybrid autoencoder (HAE) to function as a compressor to alleviate the requirement for large memory, achieving $\mathcal{O}(0.1 t)$ at the worst case with the computing complexity of $\mathcal{O}(t)$ while accomplishing state-of-the-art performance. The decoder later recovers the exemplar data stored in the latent space, rather than in raw format. Additionally, HAE is designed for both discriminative and generative modeling, enabling classification and replay capabilities, respectively. HAE adopts the charged particle system energy minimization equations and repulsive force algorithm for the incremental embedding and distribution of new class centroids in its latent space. Our results demonstrate that AHR consistently outperforms recent baselines across multiple benchmarks while operating with the same memory/compute budgets. The source code is included in the supplementary material and will be open-sourced upon publication.
title Autoencoder-Based Hybrid Replay for Class-Incremental Learning
topic Machine Learning
url https://arxiv.org/abs/2505.05926