TY  - JOUR
AU  - Zeynali, Mahsa
AU  - Seyedarabi, Hadi
AU  - Afrouzian, Reza
PY  - 2026
DA  - 2026/03/10
TI  - Image Generation Inspired by Electroencephalography for Neuromarketing Applications Using Extracted Features from Transformer-Based Models
JO  - OBM Neurobiology
SP  - 328
VL  - 10
IS  - 01
AB  - The design of products in Neuromarketing using machine learning methods has been a continuous challenge in Computer-aided design. Previously, deep learning techniques have been applied to generate random images for domains such as furniture, fashion, and product design. However, using deep generative methods requires a large amount of data and overlooks human aspects in the design process. This paper aims to extract human perceptual factors from brain signals using a Transformer-based model and involve them in artificial intelligence-based product design. To achieve this, Electroencephalography (EEG) signals are recorded while observing product images. In the first stage, an encoder based on the Transformer architecture extracts features from raw signals. In the second stage, an Auxiliary Classifier Generative Adversarial Network (ACGAN) is trained on the extracted features to generate product images. An accuracy of 92.8% obtained from the EEG encoder signifies that the features extracted by the Transformer-based model are well distinguishable as a condition of the generator. The generated images, with an inception score of 5.22 and 78% classification accuracy, exhibit features similar to those of the original images. This approach holds promise for enhancing designer-customer communication in Neuromarketing applications, particularly in scenarios where customers may struggle to express their design preferences clearly.
SN  - 2573-4407
UR  - https://doi.org/10.21926/obm.neurobiol.2601328
DO  - 10.21926/obm.neurobiol.2601328
ID  - Zeynali2026
ER  -