New developments in analysis and statistics


Get the latest news delivered in your mailbox.

General session - New developments in analysis and statistics

Schedule: Wednesday 18th May 11:00 - 13:30 CET Virtual Room 1
Session Chair: Anne-Marie Brouwer, TNO Soesterberg, The Netherlands

11:00-11:15    Collaborative learning interactions among university students in face-to-face and online meetings during the COVID-19 pandemic: An observational study.
H.Q. Chim, Mirjam G.A. Egbrink, Diana Dolmans, & Renate de Groot Renate. Maastricht University, The Netherlands.
This study explores patterns in interactions among university students learning collaboratively online and face-to-face. Observations were made in tutorials from 2018 (online), 2020 (face-to-face), and 2021 (face-to-face), on the same academic topic.

11:15-11:30    The Effects of Stimulus Duration and Group Size on Wearable Physiological Synchrony.
Ivo Stuldreher. TNO Soesterberg, The Netherlands
Individuals sharing attention to narrative stimuli show synchronized heart rate and electrodermal activity. In the current work, we investigate how significance of such inter-subject correlations depends on stimulus length and participant group size.

11:30-11:45    Start Making Sense: Predicting confidence in virtual human interactions using biometric signals.
Sara Dalzel-Job & R.L. Hill. School of Informatics, University of Edinburgh, Scotland, UK. and R. Petrick, School of Mathematical & Computer Sciences, Heriot-Watt University, Edinburgh, Scotland, UK
This project investigates the use of biometric data to predict confidence levels during task-focused interaction between humans and virtual humans. The project comprises of two main studies, the first of which examines the relationship between biometric signals – galvanic skin response (GSR), heart rate, facial expression and eye movements – and self-report levels of confidence during a task-oriented interaction between a human and a virtual human.

12:00 - 13:00 Break

13:00-13:15    Improving the Annotation Efficiency for Animal Activity Recognition using Active Learning.
Suzanne Spink, J.W. Kamminga & A. Kamilaris. University of Twente, The Netherlands
Machine learning (ML) can be used to efficiently annotate IMU animal activity data. Five variations of active learning, a semi-supervised ML algorithm, have been analysed and compared to random sampling and manual labelling.

13:15-13:30    Data Synchronisation and Processing in Multimodal Research.
Tenzing Dolmans. M. Poel, J.W.J.R van ’t Klooster, and B.P. Veldkamp, University of Twente, The Netherlands
Gathering physiological information with the help of various devices can help gain insight into mental states and constructs, such as mental workload. These devices measure, for example, galvanic skin response, heart rate, or neural activity. However, challenges arise during the collection and synchronisation when combining multiple devices to evaluate the mental state of the participant. Devices use different timestamps, communicate only with certain hardware or software, and do not always provide reliable avenues of data synchronisation after recording. In this paper, we discuss different types of approaches to a workflow and pipeline that can help overcome these problems when identifying mental workload.