Approximation theory · Neural networks
Sharp closure thresholds for two-hidden-layer ReLU networks
Sharp width thresholds for approximation by two-hidden-layer ReLU networks.
School of Artificial Intelligence
Capital University of Economics and Business
Research
Papers, preprints, and reproducibility materials.
Approximation theory · Neural networks
Sharp width thresholds for approximation by two-hidden-layer ReLU networks.
Finite geometry · Rank-metric codes
The classification of maximum scattered linear sets in PG(1,q⁵).
Discrete geometry · Quantitative stability
Coordinate balancing, equality certificates, and sharp stability for symmetric lattice cross-covariograms.