Data-Efficient Learning

Ask about this topic

Overview

Data-efficient learning studies how to build useful models when labels, examples, simulations, or interaction data are costly or limited. Common approaches include active learning and coreset selection, semi-supervised and self-supervised pretraining, transfer and meta-learning, and curricula that prioritize examples suited to a model’s current abilities. In scientific and clinical settings, physical laws, domain knowledge, and synthetic data can provide useful structure when observations are scarce. A central challenge is that the best strategy depends on the task, data distribution, and compute budget, so gains in accuracy must be weighed against training cost, robustness, and reliable evaluation.

Latest papers

CardsList

245 papers28 papers in the last four weeks, up 56% on the four weeks before. 0.2% of all new papers.

  1. Efficient Reinforcement Finetuning via Adaptive Curriculum Learning

    Apr 7, 2025Taiwei Shi, Yiyang Wu, Linxin Song +2RL for Language Model ReasoningMathematical Reasoning

    University of Southern California · Carnegie Mellon University · Mohamed Bin Zayed University of Artificial Intelligence

  2. Scaling Online Complex Event Detection with Synthetic Supervision and Mamba-Based Neural Algorithmic Reasoning

    Feb 11, 2025Liying Han, Gaofeng Dong, Xiaomin Ouyang +4Neural Algorithmic ReasoningAlgorithmic Reasoning

    Department of Electrical and Computer Engineering University of California, Los Angeles · Department of Computer Science and Engineering Hong Kong University of Science and Technology · U.S. Army DEVCOM Army Research Laboratory +1

  3. Selective Data Expansion for Model Performance

    Dec 4, 2024Jahid Hasan, Romila PradhanAlgorithmic FairnessData-Efficient Learning

    Purdue University, West Lafayette, USA

  4. Entropic perspective on AI resilience, data-minimal learning and optimal control

    Oct 19, 2023Illia HorenkoData-Efficient LearningAdversarial Robustness

    Chair for Mathematics of AI, Faculty of Mathematics, RPTU Kaiserslautern-Landau, Gottlieb-Daimler-Str. 48, Kaiserslautern, 67663, Germany.

  5. StoCFL: A Stochastically Clustered Federated Learning Framework for Non-IID Data with Dynamic Client Participation

    Mar 2, 2023Dun Zeng, Xiangjing Hu, Shiyu Liu +3Non-IID Federated LearningClustered Federated Learning

    University of Electronic Science and Technology of China · Peng Cheng Lab · Harbin Institute of Technology, Shenzhen +1