Neural network interactions that generalize follow a decay-shaped distribution over complexity, while non-generalizing interactions follow a spindle-shaped distribution, which a four-parameter fit can separate.
These findings suggest that interactions act as primitive inference patterns encoded by DNNs, forming the theoretical foundation of interaction-based theoretical frameworks
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
method 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
REJECT 1roles
method 1polarities
use method 1representative citing papers
citing papers explorer
-
Revisiting Generalization Power of a DNN in Terms of Symbolic Interactions
Neural network interactions that generalize follow a decay-shaped distribution over complexity, while non-generalizing interactions follow a spindle-shaped distribution, which a four-parameter fit can separate.