An Exponential Representation in the API Algorithm for Hidden Markov Models Training

In this paper, we show how an efficient ant based algorithm, called API and initially designed to perform real parameter optimization, can be adapted to the difficult problem of Hidden Markov Models training. To this aim, a transformation of the search space that preserves API’s vectorial moves is introduced. Experiments are conducted with various temporal series extracted from images.

• Type: Conference
• Year: 2005
• Where: AE
• Authors: Sebastien Aupetit, Nicolas Monmarché, Mohamed Slimane, Pierre Liardet
• Keywords: AE 2005, Ant Based Algorithm, API’s Vectorial Moves, Real Parameter Optimization