Summary of Don’t Think It Twice: Exploit Shift Invariance For Efficient Online Streaming Inference Of Cnns, by Christodoulos Kechris et al.
Don’t Think It Twice: Exploit Shift Invariance for Efficient Online Streaming Inference of CNNs
by Christodoulos Kechris, Jonathan Dan, Jose Miranda, David Atienza
First submitted to arxiv on: 6 Aug 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 A novel approach in deep learning time-series processing aims to optimize computational efficiency during inference by exploiting convolution’s shift-invariance properties. This strategy, called StreamiNNC, is designed for online streaming inference and addresses limitations of zero-padding and pooling operations commonly used in such networks. By introducing signal padding and pooling alignment, the method can be deployed effectively on real-world biomedical signal processing applications. Theoretical error upper bounds are derived for pooling during streaming, and experiments show that StreamiNNC achieves a low deviation between streaming output and normal inference (2.03-3.55% NRMSE) for three evaluated networks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary StreamiNNC is a new way to make computers learn from time-series data faster and more efficient. Usually, this type of learning requires extra calculations because the network looks at small parts of the data at a time. StreamiNNC gets rid of these extra calculations by using properties of convolutional neural networks that don’t change when you shift the data. This makes it possible to speed up the inference process without losing accuracy. The method is tested on simulated and real-world biomedical signal processing applications, showing promising results. |
Keywords
» Artificial intelligence » Alignment » Deep learning » Inference » Signal processing » Time series
