Introduction

Deception detection is a task of extreme importance concerning widespread issues of security. Studies are consistent in showing that deception judgement in all types of people is no more accurate than chance. [1,2] Therefore, when dealing with deception, a more precise approach is needed. Deception detection methods using modalities which include text, audio and face-to-face communication (e.g unconscious facial expressions) have been widely studied but often suffer from limitations such as being not widely applicable when not accompanied by a trained expert. We attempt to introduce methods of deception detection using modalities, besides the extensively studied audio-visual, where decisions are based on analysis and interpretation leading to use without prior training. We are focusing on a less explored method of electroencephalography (EEG), which is a medical imaging technique used to record the electrical activity of the brain. Our goal is to develop models using EEG data for binary classification of deception and non-deception groups. For creating the model different methods were considered:

• logistic regression models,
• linear and quadratic classifiers (LDA, QDA),
• decision trees,
• random forests.

Our end goal is a model with high accuracy of classification.

[1] https://firstmonday.org/ojs/index.php/fm/article/view/3933/3170
[2] https://www.ncbi.nlm.nih.gov/pubmed/16859438