AI RESEARCH
Partition of Unity Neural Networks for Interpretable Classification with Explicit Class Regions
arXiv CS.LG
•
ArXi:2602.00511v2 Announce Type: replace Despite their empirical success, neural network classifiers remain difficult to interpret. In softmax-based models, class regions are defined implicitly as solutions to systems of inequalities among logits, making them difficult to extract and visualize. We PUNN constructs $k$ nonnegative functions $h_1, \ldots, h_k$ satisfying $\sum_i h_i(x) = 1$, where each $h_i(x)$ directly represents $P(\text{class } i \mid x