Optimal Compressing and Decompressing Digital-Ink Handwriting via Sparse Gaussian Process Regression and Dynamic Programming

In this study, we delve into the challenge of compressing and decompressing digital-ink data obtained from handwritten inputs on devices such as pen tablets, etc. We present a new approach using sparse Gaussian process regression in order to compress digital-ink data into some kernel matrix consisting of a sequence of pseudo-inputs. Moreover, to enhance the accuracy of compression and decompression, we introduce a dynamic programming (DP) approach for the optimal selection of dominant pseudo-inputs. The effectiveness of our method is substantiated through a series of experimental studies.