Hacking Task Confounder in Meta-Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wang, Jingyao, Ren, Yi, Song, Zeen, Zhang, Jianqi, Zheng, Changwen, Qiang, Wenwen
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909212498460672
author Wang, Jingyao
Ren, Yi
Song, Zeen
Zhang, Jianqi
Zheng, Changwen
Qiang, Wenwen
author_facet Wang, Jingyao
Ren, Yi
Song, Zeen
Zhang, Jianqi
Zheng, Changwen
Qiang, Wenwen
contents Meta-learning enables rapid generalization to new tasks by learning knowledge from various tasks. It is intuitively assumed that as the training progresses, a model will acquire richer knowledge, leading to better generalization performance. However, our experiments reveal an unexpected result: there is negative knowledge transfer between tasks, affecting generalization performance. To explain this phenomenon, we conduct Structural Causal Models (SCMs) for causal analysis. Our investigation uncovers the presence of spurious correlations between task-specific causal factors and labels in meta-learning. Furthermore, the confounding factors differ across different batches. We refer to these confounding factors as "Task Confounders". Based on these findings, we propose a plug-and-play Meta-learning Causal Representation Learner (MetaCRL) to eliminate task confounders. It encodes decoupled generating factors from multiple tasks and utilizes an invariant-based bi-level optimization mechanism to ensure their causality for meta-learning. Extensive experiments on various benchmark datasets demonstrate that our work achieves state-of-the-art (SOTA) performance.
format Preprint
id arxiv_https___arxiv_org_abs_2312_05771
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Hacking Task Confounder in Meta-Learning
Wang, Jingyao
Ren, Yi
Song, Zeen
Zhang, Jianqi
Zheng, Changwen
Qiang, Wenwen
Machine Learning
Meta-learning enables rapid generalization to new tasks by learning knowledge from various tasks. It is intuitively assumed that as the training progresses, a model will acquire richer knowledge, leading to better generalization performance. However, our experiments reveal an unexpected result: there is negative knowledge transfer between tasks, affecting generalization performance. To explain this phenomenon, we conduct Structural Causal Models (SCMs) for causal analysis. Our investigation uncovers the presence of spurious correlations between task-specific causal factors and labels in meta-learning. Furthermore, the confounding factors differ across different batches. We refer to these confounding factors as "Task Confounders". Based on these findings, we propose a plug-and-play Meta-learning Causal Representation Learner (MetaCRL) to eliminate task confounders. It encodes decoupled generating factors from multiple tasks and utilizes an invariant-based bi-level optimization mechanism to ensure their causality for meta-learning. Extensive experiments on various benchmark datasets demonstrate that our work achieves state-of-the-art (SOTA) performance.
title Hacking Task Confounder in Meta-Learning
topic Machine Learning
url https://arxiv.org/abs/2312.05771