Probabilistic Attention for Interactive Segmentation

TL;DR

This paper introduces a probabilistic interpretation of attention in transformers, enabling online adaptation through EM algorithms, which improves interactive segmentation performance by effectively propagating user feedback.

Contribution

It presents a novel probabilistic framework for attention, allowing online adaptation of model parameters for interactive tasks, enhancing annotation efficiency.

Findings

Key adaptation improves performance by ~10% mIoU in low feedback scenarios.

Value propagation enhances responsiveness in high feedback regimes.

Probabilistic attention outperforms standard methods on benchmark datasets.

Abstract

We provide a probabilistic interpretation of attention and show that the standard dot-product attention in transformers is a special case of Maximum A Posteriori (MAP) inference. The proposed approach suggests the use of Expectation Maximization algorithms for online adaptation of key and value model parameters. This approach is useful for cases in which external agents, e.g., annotators, provide inference-time information about the correct values of some tokens, e.g, the semantic category of some pixels, and we need for this new information to propagate to other tokens in a principled manner. We illustrate the approach on an interactive semantic segmentation task in which annotators and models collaborate online to improve annotation efficiency. Using standard benchmarks, we observe that key adaptation boosts model performance ($\sim10\%$ mIoU) in the low feedback regime and value propagation improves model responsiveness in the high feedback regime. A PyTorch layer implementation of our probabilistic attention model will be made publicly available here: https://github.com/apple/ml-probabilistic-attention.