LGFeb 10, 2024
Understanding Test-Time Augmentation
arXiv:2402.06892v164 citationsh-index: 7
Originality Synthesis-oriented
AI Analysis
This work addresses a theoretical gap for researchers and practitioners using TTA in machine learning, but it is incremental as it builds on an existing heuristic.
The paper tackles the lack of theoretical understanding of Test-Time Augmentation (TTA), a heuristic that uses data augmentation during testing to improve predictions, by providing theoretical guarantees and clarifying its behavior.
Test-Time Augmentation (TTA) is a very powerful heuristic that takes advantage of data augmentation during testing to produce averaged output. Despite the experimental effectiveness of TTA, there is insufficient discussion of its theoretical aspects. In this paper, we aim to give theoretical guarantees for TTA and clarify its behavior.