Summary of [re] Network Deconvolution, by Rochana R. Obadage et al.
[Re] Network Deconvolution
by Rochana R. Obadage, Kumushini Thennakoon, Sarah M. Rajtmajer, Jian Wu
First submitted to arxiv on: 2 Oct 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Digital Libraries (cs.DL); 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 Our research aims to replicate the findings published in “Network Deconvolution” by Ye et al. (2020). The original paper proposes an optimization technique for model training in convolutional neural networks, using a method called “network deconvolution” to remove correlations before feeding data into each layer. We investigate whether using network deconvolution instead of batch normalization improves deep learning model performance. Our study confirms the validity of this claim by successfully reproducing the results reported in Tables 1 and 2 of the original paper, involving 367 unique experiments across multiple architectures, datasets, and hyperparameter configurations. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Our work is about verifying the results from a previous study on how to improve deep learning model performance. The original study used a technique called “network deconvolution” to help models learn better. We tried it out and found that it really works! We did 367 different experiments using different models, data sets, and settings, and our results matched the original study’s findings almost perfectly. |
Keywords
» Artificial intelligence » Batch normalization » Deep learning » Hyperparameter » Optimization
