Summary of Local Transfer Learning Gaussian Process Modeling, with Applications to Surrogate Modeling Of Expensive Computer Simulators, by Xinming Wang et al.
Local transfer learning Gaussian process modeling, with applications to surrogate modeling of expensive computer simulators
by Xinming Wang, Simon Mak, John Miller, Jianguo Wu
First submitted to arxiv on: 16 Oct 2024
Categories
- Main: Machine Learning (stat.ML)
- Secondary: Machine Learning (cs.LG); Applications (stat.AP); Methodology (stat.ME)
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 LOcal transfer Learning Gaussian Process (LOL-GP) model leverages a carefully-designed Gaussian process to transfer information from “source” systems to “target” systems, addressing the critical limitation of “negative transfer” in existing transfer learning models. The LOL-GP’s latent regularization model identifies regions where transfer is beneficial and regions where it should be avoided, allowing for local transfer. This approach improves surrogate performance over existing methods, as demonstrated through numerical experiments and an application in jet turbine design. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A new kind of computer model helps scientists predict how complex systems will behave. These models are trained using data from similar systems and can make accurate predictions about the behavior of a target system. The problem is that sometimes this model actually makes things worse, because it’s transferring bad information from the source system. This new model, called LOL-GP, solves this problem by only transferring good information to the target system. It does this by identifying when the source and target systems behave similarly or differently, and adjusting its transfer accordingly. This leads to more accurate predictions and is useful for things like designing new jet turbines. |
Keywords
» Artificial intelligence » Regularization » Transfer learning
