HMLET – High-resolution Machine Learning Eye-tracking Toolbox

Eye-tracking technology yields high temporal and spatial resolution data on visual behavior. Detection and quantification of cognitive processes underlying looking behavior can be challenging using conventional processing and statistical analysis. Non-parametric statistical methods are powerful alternatives for high-resolution time series analysis and data driven machine learning techniques can provide an unbiased and explorative description in the analysis of high dimensional data. This paper introduces high-resolution machine learning eye-tracking toolbox (HMLET) that provides an accurate implementation of non-parametric statistical methods in R language and an executable graphic user interface (GUI) comprising novel methods and tools for extraction and analysis of informative measures from eye-tracking data with high temporal resolution. This package supports processing of raw eye-tracking data for basic preprocessing, high-resolution time series analysis, non-parametric permutation tests, data-driven evaluation and selection of eye-tracking measures, a novel visual exploration similarity analysis, along with a sets of visualization tools. HMLET is freely available for download and contribution on GitHub https://github.com/Alireza-Kazemi/MLET, distributed under the MIT license.