Summary of Second-order Forward-mode Automatic Differentiation For Optimization, by Adam D. Cobb et al.
Second-Order Forward-Mode Automatic Differentiation for Optimization
by Adam D. Cobb, Atılım Güneş Baydin, Barak A. Pearlmutter, Susmit Jha
First submitted to arxiv on: 19 Aug 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Machine Learning (stat.ML)
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 The paper introduces a novel optimization algorithm called FoMoH (Forward-mode weight perturbation with Hessian information), which generalizes second-order line search to k-dimensional hyperplanes. By combining the forward-mode stochastic gradient method with a second-order hyperplane search, FoMoH avoids storage overhead and uses hyper-dual numbers to jointly evaluate directional derivatives and their quadratic terms. This algorithm is showcased as a potential solution for optimizing machine learning models without backpropagation. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary FoMoH is a new way to optimize machine learning models. Normally, we use something called “backpropagation” to do this, but sometimes that’s not possible or efficient. FoMoH finds a middle ground by using special numbers and math to help find the best settings for our models. This can be helpful in situations where backpropagation isn’t working well. The authors of the paper made their code available so others can try it out. |
Keywords
» Artificial intelligence » Backpropagation » Machine learning » Optimization
