---
title: "Beyond Scaling: A Survey of Data-Efficient Learning for LLM Agents"
authors:
  - "Yaqing Wang"
  - "Zhenlin Luo"
  - "Peiyao Zhao"
  - "Yunfeng Cai"
  - "Quanming Yao"
doi: "10.24963/ijcai.2026/893"
url: "https://ijcai.org/proceedings/2026/893"
pdf: "https://ijcai.org/proceedings/2026/0893.pdf"
venue: "Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence"
year: 2026
pages: "8050-8059"
track: "Survey Track"
keywords:
  - "Machine Learning: Learnware/model reuse/transfer learning"
  - "Machine Learning: Few-shot learning"
  - "Agent-based and Multi-agent Systems: Applications"
---

# Beyond Scaling: A Survey of Data-Efficient Learning for LLM Agents

Yaqing Wang, Zhenlin Luo, Peiyao Zhao, Yunfeng Cai, Quanming Yao

*Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence*, pages 8050-8059. DOI: [10.24963/ijcai.2026/893](https://doi.org/10.24963/ijcai.2026/893)

## Abstract

LLM-based agents perceive, reason, act, and adapt through interaction. While scaling remains important, agentic progress also depends on extracting more learning signal from limited experience, including demonstrations, feedback, reasoning traces, tool-use records, and interaction trajectories. This survey develops an agent-centric view of data-efficient learning. We formulate an agentic learning loop, define data efficiency as capability improvement without proportional increases in supervision or real-environment trial-and-error, and synthesize methods into experience augmentation, agent structural design, and learning paradigms. We also summarize representative application domains and open challenges in experience reuse and cost-aware evaluation. Overall, the agentic era requires smarter ways to acquire, structure, reuse, and learn from limited experience.