Akademik Veri Yönetim Sistemi
MULTIMEDIA TOOLS AND APPLICATIONS, cilt.83, sa.14, ss.40753-40771, 2024 (SCI-Expanded)
Neuromarketing relies on brain-computer interface technology to understand consumer preferences for products and services. Marketers spend approximately 400 billion dollars each year on advertising and promotion in traditional marketing. Traditional marketing approaches cannot fully explain or capture consumers' real-time decision-making. On the other hand, neuromarketing promises to get around these limitations. In this study, we presented a multi-channel electroencephalography (EEG)-based deep learning approach to classify consumers' preferences. The EEG signals were recorded from 25 subjects using 14 channels. The channels were categorized according to the frontal, parietal, temporal, and occipital brain regions. The multitaper spectral analysis approach was then used to extract the feature vectors. Using the extracted feature vectors, the performances of bidirectional long-short-term memory (Bidirectional-LSTM) deep learning, support vector machine (SVM), and k-nearest neighbors (k-NN) machine learning algorithms were compared. The performance of the algorithms was analyzed using frontal, central, parietal, temporal, and occipital brain regions and all channels. Bidirectional-LSTM deep learning algorithm attained the highest accuracy among the other experiments. According to the placement of the channels in the brain regions, the highest accuracy value was 96.83% using Bidirectional-LSTM deep learning algorithm and this was achieved by using electrodes in the frontal region. The performance results analysis was found to be 0.99 recall, 0.95 precision, 0.94 specificity, and 0.97 f1-measure. As a result, this study offers proof of deep learning algorithms' effectiveness in neuromarketing applications.