Decoding Visual Art Preferences From EEG Signals Using Wavelet Scattering Transform

Maria Kyrou, Nikos Laskaris, Panagiotis C. Petrantonakis, Fotis P. Kalaganis, Kostas Georgiadis, Spiros Nikolopoulos, Ioannis Kompatsiaris

Abstract: Understanding art preferences through neural signals can enhance artistic experiences and provide valuable insights into aesthetic perception. In this study, we propose a novel EEG-based framework for visual art preferences classification, leveraging Wavelet Scattering Transform (WST) for feature extraction and Support Vector Machines (SVM) for classification. Unlike deep learning approaches that require large-scale datasets and extensive training, a wavelet scattering network provides low-variance, translation-invariant features without the need for learnable parameters, making it well-suited for regular size EEG datasets. Experimental results demonstrate that the proposed method effectively differentiates between “like” and “dislike” ratings based on EEG responses to visual art stimuli. The findings highlight the potential of wavelet scattering-based feature extraction in decoding aesthetic preferences.

External IDs:dblp:conf/icdsp/KyrouLPKGNK25