Output details

A user-specific Machine Learning approach for improving touch accuracy on mobile devices

<20>This paper introduces a probabilistic method for learning individual offset models for mobile touchscreens, demonstrating that individual-specific models can improve button targeting performance by >20%. In addition, unique prototype hardware supplied by Nokia (seppo.turunen@nokia.com) allowed us to be the first researchers to compare offset models that use the phone-supplied coordinates as an input with those that use the raw output from the sensor array. This work has led to further collaborations with Nokia including research funding and access to state-of-the-art unreleased hardware. This work was published in ACM UIST (the top HCI venue for interactive systems technology, acceptance rate 20%).