# Enabling On-Device Learning via Experience Replay with Efficient Dataset Condensation

arXiv: https://arxiv.org/abs/2405.16113 · HTML: https://mlsanity.com/paper/2405.16113/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 20 of its 39 references that are in the map.
- First posted on arXiv 25 May 2024.

## Builds on (most cited first, 20 of 20)

- [ImageNet Large Scale Visual Recognition Challenge](https://mlsanity.com/paper/1409.0575/) (2014)
- [iCaRL: Incremental Classifier and Representation Learning](https://mlsanity.com/paper/1611.07725/) (2016)
- [DARTS: Differentiable Architecture Search](https://mlsanity.com/paper/1806.09055/) (2018)
- [Contrastive Multiview Coding](https://mlsanity.com/paper/1906.05849/) (2019)
- [Dark Experience for General Continual Learning: a Strong, Simple Baseline](https://mlsanity.com/paper/2004.07211/) (2020)
- [Gradient based sample selection for online continual learning](https://mlsanity.com/paper/1903.08671/) (2019)
- [Dataset Condensation with Gradient Matching](https://mlsanity.com/paper/2006.05929/) (2020)
- [Dataset Distillation by Matching Training Trajectories](https://mlsanity.com/paper/2203.11932/) (2022)
- [Dynamic Few-Shot Visual Learning without Forgetting](https://mlsanity.com/paper/1804.09458/) (2018)
- [GRAD-MATCH: Gradient Matching based Data Subset Selection for Efficient Deep Model Training](https://mlsanity.com/paper/2103.00123/) (2021)
- [Dataset Condensation via Efficient Synthetic-Data Parameterization](https://mlsanity.com/paper/2205.14959/) (2022)
- [Continual Prototype Evolution: Learning Online from Non-Stationary Data Streams](https://mlsanity.com/paper/2009.00919/) (2020)
- [Memory Efficient Experience Replay for Streaming Learning](https://mlsanity.com/paper/1809.05922/) (2018)
- [Towards Lossless Dataset Distillation via Difficulty-Aligned Trajectory Matching](https://mlsanity.com/paper/2310.05773/) (2023)
- [Condensing Graphs via One-Step Gradient Matching](https://mlsanity.com/paper/2206.07746/) (2022)
- [Accelerating Deep Learning by Focusing on the Biggest Losers](https://mlsanity.com/paper/1910.00762/) (2019)
- [Sample Condensation in Online Continual Learning](https://mlsanity.com/paper/2206.11849/) (2022)
- [Catastrophic Forgetting in Deep Learning: A Comprehensive Taxonomy](https://mlsanity.com/paper/2312.10549/) (2023)
- [Enabling On-Device Self-Supervised Contrastive Learning With Selective Data Contrast](https://mlsanity.com/paper/2106.03796/) (2021)
- [A Survey of Deep Learning Techniques for Mobile Robot Applications](https://mlsanity.com/paper/1803.07608/) (2018)

## Abstract (arXiv)

Upon deployment to edge devices, it is often desirable for a model to further learn from streaming data to improve accuracy. However, extracting representative features from such data is challenging because it is typically unlabeled, non-independent and identically distributed (non-i.i.d), and is seen only once. To mitigate this issue, a common strategy is to maintain a small data buffer on the edge device to hold the most representative data for further learning. As most data is either never stored or quickly discarded, identifying the most representative data to avoid significant information loss becomes critical. In this paper, we propose an on-device framework that addresses this issue by condensing incoming data into more informative samples. Specifically, to effectively handle unlabeled incoming data, we propose a pseudo-labeling technique designed for unlabeled on-device learning environments. Additionally, we develop a dataset condensation technique that only requires little computation resources. To counteract the effects of noisy labels during the condensation process, we further utilize a contrastive learning objective to improve the purity of class data within the buffer. Our empirical results indicate substantial improvements over existing methods, particularly when buffer capacity is severely restricted. For instance, with a buffer capacity of just one sample per class, our method achieves an accuracy that outperforms the best existing baseline by 58.4% on the CIFAR-10 dataset.