TU Berlin

Methoden der Künstlichen IntelligenzBachelorarbeiten

Inhalt des Dokuments

zur Navigation

Bachelor Thesis Proposition

The paper Augment and Reduce: Stochastic Inference for Large Categorical Distributions proposes to simplify multiclass classification for a large number of classes. The method proposed is based on probabilities but only aim at solving a maximum likelihood problem, the project would be to enlarge the scope of the algorithm with a prior by doing maximum-a-posteriori or maybe variational inference.

Please contact Theo Galy-Fajou if you are interested

Paper : https://arxiv.org/abs/1802.04220

Bachelorarbeiten

Parameter Estimation in Potts Models as an Instance of Probabilistic Programming via Imperatively Defined Factor Graphs
Zitatschlüssel Hanefeld:2014:PEP
Autor Alec Hanefeld
Jahr 2014
Monat July
Schule TU Berlin
Typ der Publikation Bachelorarbeit
Link zur Publikation Download Bibtex Eintrag

Navigation

Direktzugang

Schnellnavigation zur Seite über Nummerneingabe