Supervised training of conversive hidden non-markovian models: increasing usability for gesture recognition

Hidden non-Markovian Models (HnMMs) were introduced and formalized as an extension of Hidden Markov Models for the analysis of partially observable stochastic processes. Their main advantage over HMM is the possibility to model arbitrary distributions for state transition duration, so that the unobservable stochastic process needs not to be Markovian. Besides academic examples, HnMMs were applied to gesture recognition and performed well in distinguishing similar gestures with different execution speeds. While the Proxel-Method enabled the evaluation for arbitrary HnMMs, there was no opportunity to train these models. Therefore, the models for different gestures had to be parameterized manually. This fact reduced the applicability in real gesture recognition dramatically. This paper presents a solution to this problem, introducing a supervised training approach that increases the applicability of HnMMs in gesture recognition.