Topological summaries of 2D loss landscape projections, saddle counts and average persistence, correlate with accuracy and Hessian metrics for ResNets and PINNs.
Efficient k-nearest neighbor graph construction for generic similarity measures
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Evaluating Loss Landscapes from a Topology Perspective
Topological summaries of 2D loss landscape projections, saddle counts and average persistence, correlate with accuracy and Hessian metrics for ResNets and PINNs.