# Class-Incremental Exemplar Compression for Class-Incremental Learning

arXiv: https://arxiv.org/abs/2303.14042 · HTML: https://mlsanity.com/paper/2303.14042/

## Key facts

- Cited by 50 papers in the map (data through 14 Oct 2026).
- Builds on 36 of its 52 references that are in the map.
- First posted on arXiv 24 Mar 2023.

## Builds on (most cited first, 30 of 36)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization](https://mlsanity.com/paper/1610.02391/) (2016)
- [SGDR: Stochastic Gradient Descent with Warm Restarts](https://mlsanity.com/paper/1608.03983/) (2016)
- [Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks](https://mlsanity.com/paper/1703.03400/) (2017)
- [Overcoming catastrophic forgetting in neural networks](https://mlsanity.com/paper/1612.00796/) (2016)
- [Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification](https://mlsanity.com/paper/1502.01852/) (2015)
- [Learning without Forgetting](https://mlsanity.com/paper/1606.09282/) (2016)
- [Learning Deep Features for Discriminative Localization](https://mlsanity.com/paper/1512.04150/) (2015)
- [iCaRL: Incremental Classifier and Representation Learning](https://mlsanity.com/paper/1611.07725/) (2016)
- [YOLOv4: Optimal Speed and Accuracy of Object Detection](https://mlsanity.com/paper/2004.10934/) (2020)
- [Progressive Neural Networks](https://mlsanity.com/paper/1606.04671/) (2016)
- [Large Scale Incremental Learning](https://mlsanity.com/paper/1905.13260/) (2019)
- [Dataset Distillation by Matching Training Trajectories](https://mlsanity.com/paper/2203.11932/) (2022)
- [Gradient-based Hyperparameter Optimization through Reversible Learning](https://mlsanity.com/paper/1502.03492/) (2015)
- [DER: Dynamically Expandable Representation for Class Incremental Learning](https://mlsanity.com/paper/2103.16788/) (2021)
- [Meta-Transfer Learning for Few-Shot Learning](https://mlsanity.com/paper/1812.02391/) (2018)
- [Maintaining Discrimination and Fairness in Class Incremental Learning](https://mlsanity.com/paper/1911.07053/) (2019)
- [Learning without Memorizing](https://mlsanity.com/paper/1811.08051/) (2018)
- [Causal Intervention for Weakly-Supervised Semantic Segmentation](https://mlsanity.com/paper/2009.12547/) (2020)
- [FOSTER: Feature Boosting and Compression for Class-Incremental Learning](https://mlsanity.com/paper/2204.04662/) (2022)
- [Mnemonics Training: Multi-Class Incremental Learning without Forgetting](https://mlsanity.com/paper/2002.10211/) (2020)
- [Reinforced Continual Learning](https://mlsanity.com/paper/1805.12369/) (2018)
- [Adaptive Aggregation Networks for Class-Incremental Learning](https://mlsanity.com/paper/2010.05063/) (2020)
- [Anti-Adversarially Manipulated Attributions for Weakly and Semi-Supervised Semantic Segmentation](https://mlsanity.com/paper/2103.08896/) (2021)
- [A Review on Bilevel Optimization: From Classical to Evolutionary Approaches and Applications](https://mlsanity.com/paper/1705.06270/) (2017)
- [Conditional Channel Gated Networks for Task-Aware Continual Learning](https://mlsanity.com/paper/2004.00070/) (2020)
- [On Learning the Geodesic Path for Incremental Learning](https://mlsanity.com/paper/2104.08572/) (2021)
- [Memory Replay with Data Compression for Continual Learning](https://mlsanity.com/paper/2202.06592/) (2022)
- [Class Re-Activation Maps for Weakly-Supervised Semantic Segmentation](https://mlsanity.com/paper/2203.00962/) (2022)

## Built on by (most cited first, 30 of 50)

- [Learning without Forgetting for Vision-Language Models](https://mlsanity.com/paper/2305.19270/) (2023)
- [A Comprehensive Survey of Forgetting in Deep Learning Beyond Continual Learning](https://mlsanity.com/paper/2307.09218/) (2023)
- [Class-Incremental Learning with CLIP: Adaptive Representation Adjustment and Parameter Fusion](https://mlsanity.com/paper/2407.14143/) (2024)
- [CL-LoRA: Continual Low-Rank Adaptation for Rehearsal-Free Class-Incremental Learning](https://mlsanity.com/paper/2505.24816/) (2025)
- [Long-Tailed Continual Learning For Visual Food Recognition](https://mlsanity.com/paper/2307.00183/) (2023)
- [Integrating Task-Specific and Universal Adapters for Pre-Trained Model-based Class-Incremental Learning](https://mlsanity.com/paper/2508.08165/) (2025)
- [On the Effectiveness of LayerNorm Tuning for Continual Learning in Vision Transformers](https://mlsanity.com/paper/2308.09610/) (2023)
- [Wakening Past Concepts without Past Data: Class-Incremental Learning from Online Placebos](https://mlsanity.com/paper/2310.16115/) (2023)
- [Mixture of Noise for Pre-Trained Model-Based Class-Incremental Learning](https://mlsanity.com/paper/2509.16738/) (2025)
- [Bayesian Learning-driven Prototypical Contrastive Loss for Class-Incremental Learning](https://mlsanity.com/paper/2405.11067/) (2024)
- [Vector Quantization Prompting for Continual Learning](https://mlsanity.com/paper/2410.20444/) (2024)
- [External Knowledge Injection for CLIP-Based Class-Incremental Learning](https://mlsanity.com/paper/2503.08510/) (2025)
- [How to Teach Large Multimodal Models New Skills](https://mlsanity.com/paper/2510.08564/) (2025)
- [Masked Autoencoders are Efficient Class Incremental Learners](https://mlsanity.com/paper/2308.12510/) (2023)
- [iNeMo: Incremental Neural Mesh Models for Robust Class-Incremental Learning](https://mlsanity.com/paper/2407.09271/) (2024)
- [Holistic Transfer: Towards Non-Disruptive Fine-Tuning with Partial Target Data](https://mlsanity.com/paper/2311.01420/) (2023)
- [OpenCIL: Benchmarking Out-of-Distribution Detection in Class-Incremental Learning](https://mlsanity.com/paper/2407.06045/) (2024)
- [Continual Audio-Visual Sound Separation](https://mlsanity.com/paper/2411.02860/) (2024)
- [Non-exemplar Class-incremental Learning by Random Auxiliary Classes Augmentation and Mixed Features](https://mlsanity.com/paper/2304.07707/) (2023)
- [Confidence Self-Calibration for Multi-Label Class-Incremental Learning](https://mlsanity.com/paper/2403.12559/) (2024)
- [Continual Learning on CLIP via Incremental Prompt Tuning with Intrinsic Textual Anchors](https://mlsanity.com/paper/2505.20680/) (2025)
- [Anti-Forgetting Adaptation for Unsupervised Person Re-identification](https://mlsanity.com/paper/2411.14695/) (2024)
- [Little by Little: Continual Learning via Incremental Mixture of Rank-1 Associative Memory Experts](https://mlsanity.com/paper/2506.21035/) (2025)
- [Continual Adversarial Defense](https://mlsanity.com/paper/2312.09481/) (2023)
- [Reconstruct before Query: Continual Missing Modality Learning with Decomposed Prompt Collaboration](https://mlsanity.com/paper/2403.11373/) (2024)
- [Step Out and Seek Around: On Warm-Start Training with Incremental Data](https://mlsanity.com/paper/2406.04484/) (2024)
- [C-LoRA: Continual Low-Rank Adaptation for Pre-trained Visual Models](https://mlsanity.com/paper/2502.17920/) (2025)
- [Hierarchical Semantic Tree Anchoring for CLIP-Based Class-Incremental Learning](https://mlsanity.com/paper/2511.15633/) (2025)
- [Prompt-Based Exemplar Super-Compression and Regeneration for Class-Incremental Learning](https://mlsanity.com/paper/2311.18266/) (2023)
- [Probing Image Compression For Class-Incremental Learning](https://mlsanity.com/paper/2403.06288/) (2024)

## Abstract (arXiv)

Exemplar-based class-incremental learning (CIL) finetunes the model with all samples of new classes but few-shot exemplars of old classes in each incremental phase, where the "few-shot" abides by the limited memory budget. In this paper, we break this "few-shot" limit based on a simple yet surprisingly effective idea: compressing exemplars by downsampling non-discriminative pixels and saving "many-shot" compressed exemplars in the memory. Without needing any manual annotation, we achieve this compression by generating 0-1 masks on discriminative pixels from class activation maps (CAM). We propose an adaptive mask generation model called class-incremental masking (CIM) to explicitly resolve two difficulties of using CAM: 1) transforming the heatmaps of CAM to 0-1 masks with an arbitrary threshold leads to a trade-off between the coverage on discriminative pixels and the quantity of exemplars, as the total memory is fixed; and 2) optimal thresholds vary for different object classes, which is particularly obvious in the dynamic environment of CIL. We optimize the CIM model alternatively with the conventional CIL model through a bilevel optimization problem. We conduct extensive experiments on high-resolution CIL benchmarks including Food-101, ImageNet-100, and ImageNet-1000, and show that using the compressed exemplars by CIM can achieve a new state-of-the-art CIL accuracy, e.g., 4.8 percentage points higher than FOSTER on 10-Phase ImageNet-1000. Our code is available at https://github.com/xfflzl/CIM-CIL.