Summary of How Can Deep Neural Networks Fail Even with Global Optima?, by Qingguang Guan
How Can Deep Neural Networks Fail Even With Global Optima?
by Qingguang Guan
First submitted to arxiv on: 23 Jul 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Numerical Analysis (math.NA)
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 explores the relationship between global optima and performance in deep neural networks. The authors investigate how different activation functions, such as ReLU, Parametric ReLU, and Sigmoid, affect the expressiveness of shallow and deep networks. They show that, despite having global optima, extremely overfitting deep networks can still fail to perform well on classification and function approximation tasks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Deep neural networks are great for solving certain problems like classification and function approximation. By adjusting weights and biases, we can build models that make accurate predictions. But what happens if our model finds the perfect solution? Does it always work well? The answer is no. In this study, researchers found a simple trick to extend shallow networks to any depth, and they also built overfitting deep networks that didn’t perform well despite being optimal. |
Keywords
» Artificial intelligence » Classification » Overfitting » Relu » Sigmoid
