Abstract:While there are many applications of ML to scientific problems that look promising, visuals can be deceiving. Using numerical analysis techniques, we rigorously quantify the accuracy, convergence rates, and generalization bounds of certain ML models applied to linear differential equations for parameter discovery or solution finding. Beyond the quantity and discretization of data, we identify that the function space of the data is critical to the generalization of the model. A similar lack of generalization is empirically demonstrated for commonly used models, including physics-specific techniques. Counterintuitively, we find that different classes of models can exhibit opposing generalization behaviors. Based on our theoretical analysis, we also introduce a new mechanistic interpretability lens on scientific models whereby Green's function representations can be extracted from the weights of black-box models. Our results inform a new cross-validation technique for measuring generalization in physical systems, which can serve as a benchmark.
From: Alejandro Queiruga [view email]
[v1]
Wed, 18 Jun 2025 07:25:09 UTC (1,269 KB)
[v2]
Mon, 9 Feb 2026 06:32:07 UTC (1,309 KB)
[v3]
Tue, 26 May 2026 04:17:17 UTC (1,284 KB)
[v4]
Sun, 5 Jul 2026 11:48:06 UTC (1,339 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。