Papers › Fidelity Isn't Accuracy: When Linearly Decodable Functions Fail to Match the Ground Truth

Fidelity Isn't Accuracy: When Linearly Decodable Functions Fail to Match the Ground Truth

13 Jun 2025arXiv:2506.12176archive 2025-07-28

Jackson Eshbaugh

Neural networks excel as function approximators, but their complexity often obscures the nature of the functions they learn. In this work, we propose the linearity score λ(f), a simple and interpretable diagnostic that quantifies how well a regression network's output can be mimicked by a linear model. Defined as the R² between the network's predictions and those of a trained linear surrogate, λ(f) offers insight into the linear decodability of the learned function. We evaluate this framework on both synthetic (y = x sin(x) + ϵ) and real-world datasets (Medical Insurance, Concrete, California Housing), using dataset-specific networks and surrogates. Our findings show that while high λ(f) scores indicate strong linear alignment, they do not necessarily imply predictive accuracy with respect to the ground truth. This underscores both the promise and the limitations of using linear surrogates to understand nonlinear model behavior, particularly in high-stakes regression tasks.

Code

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Results from the paper archive 2025-07-28

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections