paper-with-me

Papers

Sinkhorn doubly stochastic attention rank decay analysis

2026-04-09 · Michela Lapenna, Rita Fioresi, Bahman Gharesifard arxiv

The self-attention mechanism is central to the success of Transformer architectures. However, standard row-stochastic attention has been shown to suffer from significant signal degradation across layers. In particular, it can induce rank collapse, resulting in increasingly uniform token representations, as well as entropy collapse, characterized by highly concentrated attention distributions. Recent work has highlighted the benefits of doubly stochastic attention as a form of entropy regularization, promoting a more balanced attention distribution and leading to improved empirical performance. In this paper, we study rank collapse across network depth and show that doubly stochastic attention matrices normalized with Sinkhorn algorithm preserve rank more effectively than standard Softmax row-stochastic ones. As previously shown for Softmax, skip connections are crucial to mitigate rank collapse. We empirically validate this phenomenon on both sentiment analysis and image classification tasks. Moreover, we derive a theoretical bound for the pure self-attention rank decay when using Sinkhorn normalization and find that rank decays to one doubly exponentially with depth, a phenomenon that has already been shown for Softmax.

📄 PDF Abstract BibTeX arXiv:2604.07925

Code (0)

등록된 구현이 없습니다.

Tasks

Image ClassificationSentiment Analysis

Similar Papers 제목 키워드 기반

ASAP: Amortized Doubly-Stochastic Attention via Sliced Dual Projection

2026-05-13 · Huy Tran, Max Milkert, David Hyde arxiv

Doubly-stochastic attention has emerged as a transport-based alternative to row-softmax attention, with recent Transformer variants using it to reduce attention sinks and rank collapse while improving performance. In thi…

Sinkformers: Transformers with Doubly Stochastic Attention

2021-10-22 · Michael E. Sander, Pierre Ablin, Mathieu Blondel, Gabriel Peyré

Attention based models such as Transformers involve pairwise interactions between data points, modeled with a learnable attention matrix. Importantly, this attention matrix is normalized with the SoftMax operator, which …

Image Classification

Quantum Doubly Stochastic Transformers

2025-04-22 · Jannis Born, Filip Skogh, Kahn Rhrissorrakrai, Filippo Utro 외

At the core of the Transformer, the Softmax normalizes the attention matrix to be right stochastic. Previous research has shown that this often destabilizes training and that enforcing the attention matrix to be doubly s…

Inductive BiasObject Recognition

Ranking via Sinkhorn Propagation

2011-06-09 · Ryan Prescott Adams, Richard S. Zemel

It is of increasing importance to develop learning methods for ranking. In contrast to many learning objectives, however, the ranking problem presents difficulties due to the fact that the space of permutations is not sm…

Information RetrievalRetrieval

The Homogeneity Trap: Spectral Collapse in Doubly-Stochastic Deep Networks

2026-01-05 · Yizhi Liu arxiv

Doubly-stochastic matrices (DSM) are increasingly utilized in structure-preserving deep architectures -- such as Optimal Transport layers and Sinkhorn-based attention -- to enforce numerical stability and probabilistic i…