Summary of Improving Shift Invariance in Convolutional Neural Networks with Translation Invariant Polyphase Sampling, by Sourajit Saha et al.
Improving Shift Invariance in Convolutional Neural Networks with Translation Invariant Polyphase Sampling
by Sourajit Saha, Tejas Gokhale
First submitted to arxiv on: 11 Apr 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper investigates the impact of downsampling operators on the shift invariance of convolutional neural networks (CNNs). The authors find that existing downsampling operators can break CNNs’ shift invariance and propose a learnable pooling operator called Translation Invariant Polyphase Sampling (TIPS) to address this issue. TIPS is designed to reduce the maximum-sampling bias (MSB) and learn translation-invariant representations. The proposed method can be integrated into any CNN with minimal computational overhead and demonstrates consistent performance gains on multiple benchmarks for image classification and semantic segmentation. Additionally, TIPS improves adversarial and distributional robustness. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper looks at how downsampling affects the way convolutional neural networks (CNNs) work. The authors realize that existing ways of downsampling can make CNNs less good at recognizing things if they’re slightly shifted or moved. They create a new way to downsample called TIPS, which helps CNNs learn better and more consistent features. This method is easy to add to any CNN and does not require much extra computation. The results show that TIPS makes the CNNs perform better on many tasks and also makes them more robust against bad data. |
Keywords
» Artificial intelligence » Cnn » Image classification » Semantic segmentation » Translation
