跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2609.38278 · v2

Masked Swingers: Harnessing Data Augmentation to Advance Autoencoders for Self-Supervised Learning

首次提交 2026-09-29 23:47
版本更新 2026-11-18 23:45
雷达观测 2026-11-19 15:04
arXiv ↗PDF ↗

摘要

Self-supervised learning (SSL) trains on inputs alone, allowing us to make models that are more useful in more settings. The autoencoder SSL framework learns by reconstructing its own input after information loss through a bottleneck or noise injection. Masked autoencoders (MAE) are the most successful instantiation of this framework: they encode a random subset of patches, then decode the masked-out patches. In this work, we introduce key modifications to improve MAEs. Our method augments an image in two different ways, then masks and encodes each view separately. It then exchanges the global representations (CLS tokens) between views before decoding the masked patches. By design, our Masked Swingers encourages learning a view-agnostic summary of the image for efficient transfer. We perform extensive experiments, and find Masked Swingers outperforms the official MAE checkpoint by +10.7% averaged over 10 frozen-feature tasks at equal pre-training compute. Gains are highest on fine-grained tasks, e.g., instance retrieval. Swingers also reduces error by 64% relative to MAE on three new state-probing datasets for world modeling, while keeping the fine-tunability of MAE. Post-training by swinging nearest neighbors gains again, e.g., by +5% on IN-1K kNN, opening the door to learning invariances via autoencoding.

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存,登录后共享已有结果。

版本差异

本站首次收录为 v2,暂无更早版本可对照。

观测到的版本

相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。

v2 · 2026-11-18 23:45 · 最新观测版本

Masked Swingers: Harnessing Data Augmentation to Advance Autoencoders for Self-Supervised Learning

Self-supervised learning (SSL) trains on inputs alone, allowing us to make models that are more useful in more settings. The autoencoder SSL framework learns by reconstructing its own input after information loss through a bottleneck or noise injection. Masked autoencoders (MAE) are the most successful instantiation of this framework: they encode a random subset of patches, then decode the masked-out patches. In this work, we introduce key modifications to improve MAEs. Our method augments an image in two different ways, then masks and encodes each view separately. It then exchanges the global representations (CLS tokens) between views before decoding the masked patches. By design, our Masked Swingers encourages learning a view-agnostic summary of the image for efficient transfer. We perform extensive experiments, and find Masked Swingers outperforms the official MAE checkpoint by +10.7% averaged over 10 frozen-feature tasks at equal pre-training compute. Gains are highest on fine-grained tasks, e.g., instance retrieval. Swingers also reduces error by 64% relative to MAE on three new state-probing datasets for world modeling, while keeping the fine-tunability of MAE. Post-training by swinging nearest neighbors gains again, e.g., by +5% on IN-1K kNN, opening the door to learning invariances via autoencoding.

查看此版本解读 →查看此版本原文 ↗