Совместный семинар ИППИ РАН, НИУ ВШЭ и Сколтеха
1 марта (среда), 1830, аудитория 615 ИППИ РАН
Юрий Максимов (Сколтех, ИППИ РАН, Национальная лаборатория Лос Аламоса)
Multi-Class Classification: How to Deal with Multi-class Huge Scale Problems Efficiently?
Multi-Class Classification: How to Deal with Multi-class Huge Scale Problems Efficiently?
Multi-class classification problems are motivated by a numerous applications in nformational retrieval, text analysis and engineering systems. Nowadays problems require time and memory efficient algorithms with strong statistical guarantees. In our talk we focus on a line of multi-class learning strategies proposed in a series of recent works joint with MR Amini, Z Harchaoui, F. Iutzeler, B, Joshi, I Partalas and D Reshetova. Finally we illustrate the efficiency of our algorithms on several practical problems arises in text analysis and energy systems.
| 27.02.2017 | |