Data-Driven Design of HMM Topology for Online Handwriting Recognition

Although HMM is widely used for on-line handwriting recognition, there is no simple and wellestablished method of designing the HMM topology. We propose a data-driven systematic method to design HMM topology. Data samples in a single pattern class are structurally simplified into a sequence of straight-line segments, and then these simplified representations of the samples are clustered. An HMM is constructed for each of these clusters, by assigning a state to each straight-line segments. Then the resulting multiple models of the class are combined to form an architecture of a multiple parallel-path HMM, which behaves as a single HMM. To avoid excessive growing of the number of the states, parameter tying is applied in that structural similarity among patterns is reflected. Experiments on on-line Hangul recognition showed about 19 % of error reductions, compared to the intuitive design method.