Efficient representations and solutions for large decision problems with continuous and discrete variables are among the most important challenges faced by the designers of automa...
Branislav Kveton, Milos Hauskrecht, Carlos Guestri...
This paper describes our work on Bengali Part of Speech (POS) tagging using a corpus-based approach. There are several approaches for part of speech tagging. This paper deals with ...
Neural networks are a useful alternative to Gaussian mixture models for acoustic modeling; however, training multilayer networks involves a difficult, nonconvex optimization that...
Contemporary face recognition algorithms rely on precise
localization of keypoints (corner of eye, nose etc.). Unfortunately,
finding keypoints reliably and accurately remains
a...
: In this study, we introduce a set of one-dimensional features to represent two dimensional shape information for HMM (Hidden Markov Model) based handwritten optical character rec...