跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2605.17671 · v2

PEIRA: Learning Predictive Encoders through Inter-View Regressor Alignment

首次提交 2026-05-18 06:04
版本更新 2026-11-16 07:41
雷达观测 2026-11-17 17:12
arXiv ↗PDF ↗

摘要

Non-contrastive self-supervised learning (SSL) is an effective framework for predictive representation learning, but popular (and in practice effective) methods such as SimSiam, BYOL, I-JEPA or DINO, which rely on a form of self-distillation to train a teacher-student network, remain poorly understood as they typically do not minimize a well-defined objective. We analyze the dynamics of a variant of the Joint Embedding Predictive Architecture (JEPA) using a regularized linear regressor to predict the learned representations of two views of the data from one another, and fully characterize its stability: non-collapsed stable equilibria align with leading nonlinear canonical correlation subspaces, while collapsed equilibria may also be stable attractors. Motivated by this result, we introduce PEIRA, a non-contrastive SSL method with an explicit objective defined through the trace of the optimal linear regressor. We show that its only stable equilibria are nontrivial global minimizers and recover the same canonical correlation subspaces, with regularization selecting the effective dimension. Experiments on ImageNet-1K and CIFAR-10 show PEIRA is competitive with VICReg and LeJEPA baselines, and qualitative empirical results support the theory. Project page: https://peira-ssl.github.io. Code: https://github.com/MichaelArbel/peira-loss (loss implementation), https://github.com/Basile-Terv/PEIRA (experiment reproduction).

作者备注:NeurIPS 2026 camera-ready. Project page: https://peira-ssl.github.io. v2: corrected assumptions in the PEIRA stability result; new results with a wider projector (k=1024, 69.17% ImageNet-1K linear probe), transfer to 8 datasets, ablations, wall-clock cost, synthetic and initialization-scale experiments; links to code

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 · 2026-11-16 07:41 · 最新观测版本

PEIRA: Learning Predictive Encoders through Inter-View Regressor Alignment

Non-contrastive self-supervised learning (SSL) is an effective framework for predictive representation learning, but popular (and in practice effective) methods such as SimSiam, BYOL, I-JEPA or DINO, which rely on a form of self-distillation to train a teacher-student network, remain poorly understood as they typically do not minimize a well-defined objective. We analyze the dynamics of a variant of the Joint Embedding Predictive Architecture (JEPA) using a regularized linear regressor to predict the learned representations of two views of the data from one another, and fully characterize its stability: non-collapsed stable equilibria align with leading nonlinear canonical correlation subspaces, while collapsed equilibria may also be stable attractors. Motivated by this result, we introduce PEIRA, a non-contrastive SSL method with an explicit objective defined through the trace of the optimal linear regressor. We show that its only stable equilibria are nontrivial global minimizers and recover the same canonical correlation subspaces, with regularization selecting the effective dimension. Experiments on ImageNet-1K and CIFAR-10 show PEIRA is competitive with VICReg and LeJEPA baselines, and qualitative empirical results support the theory. Project page: https://peira-ssl.github.io. Code: https://github.com/MichaelArbel/peira-loss (loss implementation), https://github.com/Basile-Terv/PEIRA (experiment reproduction).

查看此版本解读 →查看此版本原文 ↗