Book Details

Internet Resource: Please Login to download book

In this project we propose an app that functions as an office helper system with a job interview ranker. To help with the hiring process, automated deception detection was performed by analyzing verbal and non-verbal cues and deploying various machine learning methods in order to detect deceit. The best results on video, audio and text were 97% , 96%, 92% respectively. Audio models used were SVM reaching 96% on the real-life dataset. Text models , SVM , NB reached 92% , 88% respectfully on Real-life dataset. Image models CNN reached 97% on real-life dataset.