EXPLAINING PHYSIOLOGICAL AFFECT RECOGNITION WITH OPTIMIZED ENSEMBLES OF CLUSTERED EXPLAINABLE MODELS

Affect recognition tasks involving physiological signals are difficult to generalize across a large population due low signal-to-noise ratio and limited data availability. In addition, the use of deep learning models makes it determine cause-and-effect between affect labeled affect. This work addresses following issues: uneven distribution noisy were addressed using K-Means-SMOTE Fuzzy ART (FA). The clustered hyper-rectangles extracted from FA topology fitted an Explainable Boosting Machines ensemble Easy Ensemble strategy. hyper parameters overall methodology tuned genetic algorithms for improved generalization. proposed method was tested three publicly available datasets: DEAP, DREAMER, AMIGOS. Step-by-step benchmarks showed that combining techniques achieved good generalization generated explainable information correlating features affective labels.