Master of Science ETH in Computer Science

Major: Machine Intelligence

Minor: Data Management

Semester Project: “3D Reconstruction and Semantic Labelling of Scenes from House-tour Videos”, supervised by ETH professor Marc Pollefeys and co-supervised by Dr. Dániel Béla Baráth and assistant professor of Stanford University Iro Armeni

Diploma in Electrical and Computer Engineering

Admission ranking: 1st

GPA: 9.22/10 “Excellent”

Area of Specialisation: Information Technology

Diploma Thesis: “Data augmentation: Testing the effectiveness of the mixup technique in affective computing tasks in-the-wild”, supervised by NTUA professor Stefanos Kollias and co-adviced by assistant professor of Queen Mary University of London Dimitrios Kollias ∼ Grade 10/10

Senior High School Diploma

GPA: 20/20 (Valedictorian)

Greek University Entrance Examination: 2nd highest nationwide marks for the Scientific field “Positive and Technological Sciences” (19,653/20,000)

IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

Description: In this paper, we examine the effectiveness of Mixup for in-the-wild FER in which data have large variations in head poses, illumination conditions, backgrounds and contexts. We then propose a new data augmentation strategy which is based on Mixup, called MixAugment. According to this, the network is trained concurrently on a combination of virtual examples and real examples; all these examples contribute to the overall loss function. We conduct an extensive experimental study that proves the effectiveness of MixAugment over Mixup and various state-of-the-art methods. We further investigate the combination of dropout with Mixup and MixAugment, as well as the combination of other data augmentation techniques with MixAugment.

Authors: Andreas Psaroudakis, Dimitrios Kollias

Paper: PDF

Presentation: VIDEO

National Technical Univeristy of Athens, Greece

Laboratory teaching assistant for the course “Neural Networks and Intelligent Systems”

Conducted research on human affect recognition tasks using Deep Learning Models

Working on both small and large-scale facial databases (e.g. RAF-DB, AffectNet) that are manually annotated for the presence of seven discrete facial expressions (categorical model)

Athens, Greece

Tutored high school students in Mathematics and Physics

Remedial teaching of Mathematics and Programming in preparation for the Nationwide University Entrance Examination