The field of affective computing and human-computer interaction is moving towards the development of more generalizable and robust models that can accurately detect and recognize human emotions, cognitive load, and other physiological signals. Recent research has focused on multimodal approaches, combining different modalities such as EEG, eye movement, facial expression, and physiological signals to improve the accuracy and reliability of emotion recognition and cognitive load detection systems. These approaches have shown promising results, with some studies achieving high accuracy rates in emotion recognition and cognitive load detection. Notably, the use of deep learning models and multimodal fusion techniques has been particularly effective in improving the performance of these systems.
Some noteworthy papers in this area include: REVELIO, which introduces a new multimodal dataset for task load detection and evaluates the performance of state-of-the-art models on multiple modalities and application domains. MuMTAffect, which presents a novel multimodal multitask affective framework for personality and emotion recognition from physiological signals. ADHDeepNet, which proposes a deep learning model for improving ADHD diagnosis precision and timeliness using raw EEG signals.