3 papers
cs.LG2026
Attractor Geometry Determines the Identifiability Limits of System Discovery
Matteo Gallo, Fabio Anselmi, Paolo Lazzari
Symbolic discovery of governing equations from data is limited not only by algorithm design and data volume, but by the geometry of the attractor: what the long-run dynamics allow…
cs.LG2026
PAC--Bayes Bounds on Quotient Parameter Spaces: Geometry-induced Implicit-Bias Priors
Nicola Aladrah, Fabio Anselmi
Overparameterized models often have continuous parameter symmetries, so different parameters define the same predictor. We show that PAC--Bayesian analysis should be performed on t…
cs.LG2026
Understanding and inverse design of implicit bias in stochastic learning: a geometric perspective
Nicola Aladrah, Emanuele Ballarin, Matteo Biagetti +3
A key challenge in machine learning is to explain how learning dynamics select among the many solutions that achieve identical loss values in overparameterized models - a phenomeno…