Temporal Functional Circuits: From Spline Plots to Faithful Explanations in KAN Forecasting
Naveen Mysore · 2026
Read Temporal Functional Circuits: From Spline Plots to Faithful Explanations in KAN Forecasting by Naveen Mysore (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-05685.txt.
Description
Kolmogorov-Arnold Networks replace fixed activations with learnable curves on every connection, which should make them explainable. Naveen Mysore builds Temporal Functional Circuits to turn those curves into faithful explanations for forecasting. The framework sits on a gated residual network that splits each forecast into a linear base and a sparse learned correction.
Each connection is mapped to the input lags it reads and ranked by its activation range. Faithfulness is then tested by intervention: zeroing edges or removing the spline parts and watching the forecast change. Explanations that survive those tests earn trust; the rest are exposed as decoration.
Catalog record
| CREATOR | Naveen Mysore |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 50 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.05685