Summary of Global Optimisation Of Black-box Functions with Generative Models in the Wasserstein Space, by Tigran Ramazyan et al.
Global Optimisation of Black-Box Functions with Generative Models in the Wasserstein Space
by Tigran Ramazyan, Mikhail Hushchyn, Denis Derkach
First submitted to arxiv on: 16 Jul 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: High Energy Physics – Experiment (hep-ex); Data Analysis, Statistics and Probability (physics.data-an); 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 proposed uncertainty estimator for gradient-free optimization of black-box simulators uses deep generative surrogate models. The challenge lies in optimizing stochastic simulators in higher dimensions. A deep generative surrogate approach is used to model the black box response, allowing for the estimation of Wasserstein uncertainty based on the Wasserstein distance. This method is incorporated into a posterior agnostic gradient-free optimization algorithm that minimizes regret over the entire parameter space. The results show that this approach is more robust than state-of-the-art methods like efficient global optimization with a deep Gaussian process surrogate. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary We’re working on optimizing black-box simulators, which are tricky to work with because they can be unpredictable and difficult to understand. To make things easier, we created a new way to estimate uncertainty using special models that can learn from data. This helps us figure out what’s most likely to happen when we try to optimize the simulator. Our approach is better than other methods at dealing with simulators that are hard to work with. |
Keywords
* Artificial intelligence * Optimization
