Bioinformatics and handwriting/speech recognition: unconventional applications of similarity search tools

This work introduces two unconventional applications for sequence alignment algorithms outside the domain of bioinformatics: handwriting recognition and speech recognition. In each application we treated data samples, such as the path of a handwritten pen stroke, as a protein sequence and use the FastA sequence alignment tool to classify unknown data samples, such as a written character. That is, we handle the handwriting and speech recognition problems like the protein annotation problem: given a sequence of unknown function, we annotate the sequence via sequence alignment. This approach achieves classification rates of 99.65% and 93.84% for the handwriting and speech recognition respectively. In addition, we provide a framework for applying sequence alignment to a variety of other non–traditional problems.