ReLU activation patterns create polytope decompositions whose dual graph Fiedler partitions correlate with decision boundaries, and cell counts track training loss.
Algorithmic determination of the combinatorial structure of the linear regions of relu neural networks.SIAM Journal on Applied Algebra and Geometry, 9(2):374–404
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Topological Signatures of ReLU Neural Network Activation Patterns
ReLU activation patterns create polytope decompositions whose dual graph Fiedler partitions correlate with decision boundaries, and cell counts track training loss.