Summary of The Local Learning Coefficient: a Singularity-aware Complexity Measure, by Edmund Lau et al.
The Local Learning Coefficient: A Singularity-Aware Complexity Measure
by Edmund Lau, Zach Furman, George Wang, Daniel Murfet, Susan Wei
First submitted to arxiv on: 23 Aug 2023
Categories
- Main: Machine Learning (stat.ML)
- Secondary: Artificial Intelligence (cs.AI); 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 introduces the Local Learning Coefficient (LLC), a novel measure to quantify the complexity of deep neural networks (DNNs). Building upon Singular Learning Theory (SLT), the LLC recognizes the significance of singularities in the loss landscape geometry. The paper provides an in-depth exploration of the LLC’s theoretical foundations, including a clear definition and intuitive insights into its application. A scalable estimator for the LLC is proposed, applied to diverse architectures, including deep linear networks, ResNet image models, and transformer language models. Empirical evidence suggests that the LLC offers valuable insights into how training heuristics influence the effective complexity of DNNs. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper looks at a new way to measure the complexity of artificial neural networks. These networks are used for tasks like recognizing pictures or understanding speech. The new method, called the Local Learning Coefficient (LLC), helps us understand how these networks work and why they might be more complex than we thought. The LLC uses some advanced math concepts to see how well these networks can learn and adapt. By testing this method on different types of neural networks, researchers found that it provides valuable insights into how these networks are influenced by the way they’re trained. |
Keywords
* Artificial intelligence * Resnet * Transformer
