Matus Telgarsky
where else to find them
no profile links on file
what they talk about
where they keep going back
no series more than once — all 5 in the filters
on a stage with
never shared a talk
who talks about the same things
nobody close enough yet
| title | conference | date |
|---|---|---|
| Modern paradigms of generalization, the heliocentric model of Aristarchus,... | Modern Paradigms in Generalization Boot Camp 2024 | 2024-08-27 |
Four concrete threads: formalize why next-token prediction yields broad downstream performance and why GPT vs BERT differ; explain LLaMA 3’s empirical trick of… | ||
| Modern Paradigms in Generalization | 9th Annual Simons Institute Industry Day | 2023-11-02 |
Program sets out to produce a taxonomy of settings and to adapt old proofs while building new tools for non IID generalization, with concrete activities: four… | ||
| Feature Selection with Gradient Descent on Two-layer Networks in Low-rotation Regimes | Deep Learning Theory Workshop and Summer School 2022 | 2022-08-01 |
Gradient flow and SGD on two‑layer ReLU nets show concrete selection and generalization claims in low‑rotation regimes: near initialization they exit the NTK… | ||
| A Primal-dual Analysis of Margin Maximization by Steepest Descent Methods | Frontiers of Deep Learning 2019 | 2019-07-15 |
Proves that steepest-descent dynamics on exponential/logistic losses drive linear predictors toward the max‑margin direction: norms diverge (roughly like log… | ||
| Approximation Power | Deep Learning Boot Camp 2019 | 2019-05-28 |
Shows that continuous univariate functions are representable as integrals of step/ReLU atoms by sampling the derivative or second derivative, and that… | ||
| nothing matches | ||