A Sparse Representation Classification Scheme for the Recognition of Affective and Cognitive Brain Processes in Neuromarketing

Abstract

In this work, we propose a novel framework to recognize the cognitive and affective processes of the brain during neuromarketing-based stimuli using EEG signals. The most crucial component of our approach is the proposed classification algorithm that is based on a sparse representation classification scheme. The basic assumption of our approach is that EEG features from a cognitive or affective process lie on a linear subspace. Hence, a test brain signal can be represented as a linear (or weighted) combination of brain signals from all classes in the training set. The class membership of the brain signals is determined by adopting the Sparse Bayesian Framework with graph-based priors over the weights of linear combination. Furthermore, the classification rule is constructed by using the residuals of linear combination. The experiments on a publicly available neuromarketing EEG dataset demonstrate the usefulness of our approach. For the two classification tasks offered by the employed dataset, namely affective state recognition and cognitive state recognition, the proposed classification scheme manages to achieve a higher classification accuracy compared to the baseline and state-of-the art methods (more than 8% improvement in classification accuracy).

Keywords: brain computer interfaces; electroencephalography; neuromarketing; sparse representation classification.

Figure 1

Averaged classification accuracy (with standard error) between the least and most preferred products.

Figure 2

Overall accuracy and confusion matrices for each method with respect to products’ preferences. Each matrix provides the overall performance of each classifier with respect to each class (in our case, product’s preferences). Furthermore, class-wise precision (last two separated columns on the right) and class-wise recall (last two separated rows on the bottom) are provided.

Figure 2

Overall accuracy and confusion matrices for each method with respect to products’ preferences. Each matrix provides the overall performance of each classifier with respect to each class (in our case, product’s preferences). Furthermore, class-wise precision (last two separated columns on the right) and class-wise recall (last two separated rows on the bottom) are provided.

Figure 3

Overall accuracy and confusion matrices for each method with respect to which product the participant views. These matrices provide the performance of each classifier with respect to each class (in our case participant views). Furthermore, the class-wise precision (last two separated columns on the right) and class-wise recall (last two separated rows on the bottom) are provided.

Figure 3

Overall accuracy and confusion matrices for each method with respect to which product the participant views. These matrices provide the performance of each classifier with respect to each class (in our case participant views). Furthermore, the class-wise precision (last two separated columns on the right) and class-wise recall (last two separated rows on the bottom) are provided.

Figure 4

Averaged classification accuracy by changing the number of training samples from 20 to 160 training samples.