The Best Paper Selection Committee (BPSC) unanimously decided to propose the paper:
Peixiang Zhong, Di Wang, and Chunyan Miao, EEG-Based Emotion Recognition Using Regularized Graph Neural Networks IEEE Trans. Aff. Comp., Vol. 13, No. 3, p.1290- 1301 July-Sept 2022.
for the IEEE Best Paper Award and the paper:
as a runner-up.
Description of the selection process and arguments in favor of the two papers:
The Best Paper Selection Committee (BPSC) of IEEE Trans. on Affective Computing (IEEE TAC) was approved by TOC chair, Jaideep Vaidya, on Jun 19th 2023 and consists of the following Associate Editors of TAC:
- Erik Cambria, National Technological University, Singapore, cambria@ntu.edu.sg
- Emily Provost (chair), University of Michigan, USA, emilykmp@umich.edu
- Albert Salah, Utrecht University, NL, a.a.salah@uu.nl
- Shangfei Wang, University of Science and Technology of China, CN, sfwang@ustc.edu.cn
The BPSC identified the following paper as the clear winner:
- Peixiang Zhong, Di Wang, and Chunyan Miao. EEG-Based Emotion Recognition Using Regularized Graph Neural Networks. IEEE Trans. Aff. Comp., Vol. 13, No. 3, p.1290-1301 July-Sept 2022
The recognition of emotional states in humans using electroencephalographic (EEG) data has seen a big increase in research interest over the past few years. EEG signals measure voltage fluctuations from the cortex in the brain and have been shown to reveal important information about human emotional states. In this paper, the authors marry EEG data with graph neural networks (GNNs), by noting that the EEG data is structured, and that this structure can be used to learn more effective classifiers. The EEG structure, comes from the brain structures it is measured from, and consist of a combination of short and long scale connections. The paper advances the state of the nnart in this field, outperforming it on ¾ experiments, and nearly matching it in the fourth. It simultaneously resolves problems of inter-subject variability and inconsistent/noisy emotion labels. However, the bigger impact of the paper is the way they integrate brain structure into their learning algorithm using a GNN. This paper, with now 290 citations on Google Scholar (156 on IEEE Explore), is garnering such support as it really opens up a new methodological area in the study of EEG emotion recognition.
As a runner up, the BPSC identified the following paper:
- Siyang Song , Shashank Jaiswal , Linlin Shen , and Michel Valstar. Spectral Representation of Behaviour Primitives for Depression Analysis IEEE Trans. Aff. Comp., Vol. 13, No. 2, pp.829-844 April-June 2022.
This paper has garnered 70 citations on Google scholar since its publication in 2020, which is considerable. The committee felt that the significance of this work lies in both the methodological contribution and the importance of the topic. Depression and mental health have become more are more visible in our society, and there is a major lack of resources for handling the increases in depression recent years. Being able to rapidly screen for depression symptoms may be a valuable medical tool that can assist in getting help to people before it is too late. Song and colleagues tackle three major problems in the automatic detection of depression symptoms. First, they handle variable length videos, a challenge for many machine learning techniques. Second, they find features that operate at multiple time scales, allowing for better integration of time. Third, they make their system context dependent, so recognition rates can be improved within given contexts. They show very promising results, and the committee believes this will help to advance research in this field in the years ahead. The committee also noted the paper is very well written and clear.