2605.30015v1 May 28, 2026 cs.LG

Test Time Training for Supervised Causal Learning

Qiang Fu
Qiang Fu
Citations: 56
h-index: 2
Rui Ding
Rui Ding
Citations: 159
h-index: 4
Dongmei Zhang
Dongmei Zhang
Citations: 1,891
h-index: 25
Jiaru Zhang
Jiaru Zhang
Citations: 278
h-index: 6
Zizhen Deng
Zizhen Deng
Citations: 5
h-index: 2
Huang Bojun
Huang Bojun
Citations: 38
h-index: 1
Shi Han
Shi Han
Citations: 2,548
h-index: 29
Jinzhuo Wang
Jinzhuo Wang
Citations: 58
h-index: 3

Supervised Causal Learning (SCL) has shown promise in causal discovery by framing it as a supervised learning problem. However, it suffers from significant out-of-distribution generalization challenges. We reveal three limitations of previous SCL practices: a significant performance gap between synthetic benchmarks and real-world data, fragility to distribution shifts, and failure in compositional generalization, collectively questioning its real-world applicability. To address this, we propose Test-Time Training for Supervised Causal Learning (TTT-SCL), a novel framework that dynamically generates training sets explicitly aligned with any specific test instance. We demonstrate the correlation between TTT-SCL and score-based methods, and design an efficient module for generating training sets based on the classic scoring function. Experiments on synthetic benchmarks, pseudo-real and real-world datasets demonstrate that TTT-SCL significantly outperforms existing SCL and traditional causal discovery methods.

0 Citations
0 Influential
14.5 Altmetric
72.5 Score
Original PDF

No Analysis Report Yet

This paper hasn't been analyzed by Gemini yet.

Log in to request an AI analysis.

댓글

댓글을 작성하려면 로그인하세요.

아직 댓글이 없습니다. 첫 번째 댓글을 남겨보세요!