Comparison of Hidden Markov Models and Support Vector Machines for vehicle crash detection

In this paper, machine-learning techniques, such as Hidden Markov Models (HMMs) and Support Vector Machines (SVMs), are applied to discriminate vehicle crashes. Both HMMs and SVMs are optimal with respect to their individual objectives. However, the performances between these two models may be very different depending on how their respective assumptions match with reality. With the exception of some benchmark reports, there is little literature comparing these two promising pattern recognition approaches. In this paper, we compare these two models analytically and experimentally, using a third intermediary method of the Linear Discriminant Analysis (LDA) for detection of crash pulses.