Laurel Heck (SZTE): Geometric and Combinatorial Tools in Machine Learning

In this talk I will introduce the discrepancy theory problem from combinatorics and its geometric generalizations, including a summary of classical results. The discrepancy theory problem has recently found many applications in diverse areas of machine learning; I will present discrepancy-based approaches and results for two different problems: coresets for kernel density estimation and quantization of large neural networks.

Az előadás helyszíne: ELTE TTK Déli tömb, harmadik emelet, D 3-316 terem (1117 Budapest, Pázmány Péter sétány 1/c)