Machine Learning for Healthcare Applications. Группа авторов

i.e., Decision Tree, Bagging Decision Tree, Gaussian Naïve Bayes Classifier, Kernel SVM and Random Forest Classifier.

Figure 3.4 Images used for visual evaluation.

Figure 3.5 Sample of EEG signal for a product with corresponding Brain map and choice label.

      We compiled all participant’s EEG data into a single file called as “Master file” with appropriate like=1/dislike=0 labelling for all rows. We observe here that Kernel SVM has the highest achieved accuracy followed by Decision Tree whereas all other 3 produce near about close results.

In Figure 3.7, we shall be showing the individual ROC AUC curve results in the graphical form for all the algorithms used on Master file showing true positive and false positive rates by its axes depicting the performance of a classification model at all classification thresholds.

Figure 3.6 Accuracy for all users (compiled).

Figure 3.7 Individual result of each algorithm.

Figure 3.8 Result of 25-users with different algorithms.

In Figure 3.8 we shall be showing the individual 25 participant’s resultant accuracy with respect to Decision Tree, Bagging Decision Tree, Gaussian naïve Bayes Classifier, Kernel SVM and Random Forest Classifier in the respective order.

      3.5.2 Comparative Results Analysis

In Figure 3.9, we shall be doing a comparative analysis of each user’s data’s performance regarding each individual algorithm in a chart form.

      In this study, we applied our knowledge of EEG data and Machine Learning to cohabit in a system for correctly analyze and predict the consumer’s choice when surveying different brands of same type of products. We had 25 males perform this initial study and it resulted in a viable feasibility for developing solutions using EEG data to enhance productivity, cut down on losses and shifting the paradigm of marketing to new heights. We have noticed that on a user-level Kernel SVM has performed better than others in majority of the cases for identifying like/dislike. It has also recorded the highest accuracy in Master file run of 56.2% among others. We have observed that Kernel SVM: Sigmoid is significant to our study and we shall try different kernels in this form to test better results.

In the following Figures 3.10 and 3.11 we shall be showing you the Approximate Brain’s EEG activity map we have derived for like/dislike states of mind in our Neuromarketing study.

Figure 3.9 Result of 25-users compared with different algorithms.

3.6 Conclusion

In our study we did not achieve high yield of results for now in predicting like/dislike of users as this is a feasibility study based on trials. Thus, we have inferred that we need to use higher complexity models of Deep Learning for achieving better accuracy in near future. We also need to work on a method to tackle fake-responses. We have future plans to provide the participants with a wide variety of choice and more neutral choices to objectively analyze the decision-making process while tracking their eye-movement as an additional parameter while delving deeper into complex classifiers for achieving higher predictions accuracy. We also think that additional methods can be added on to this process like eye-tracking coupled with higher complexity models of classifiers to take our prediction accuracy to a higher level.

Figure 3.10 Approximate brain EEG map for dislike state.

Figure 3.11 Approximate brain EEG map for like state.

References

      1. Yadava, M., Kumar, P., Saini, R., Roy, P.P., Dogra, D.P., Analysis of EEG signals and its application to neuromarketing. Multimedia Tools Appl., 76, 18, 19087–19111, 2017.

      2. Tripathi, S., Acharya, S., Sharma, R.D., Mittal, S., Bhattacharya, S., Using Deep and Convolutional Neural Networks for Accurate Emotion Classification on DEAP Dataset. Twenty-ninth IAAI conference, pp. 4746–4752, 2017.

      3. Djamal, E.C. and Lodaya, P., EEG based emotion monitoring using wavelet and learning vector quantization. 2017 4th international conference on Electrical Engineering, Computer Science and Informatics (EECSI), pp. 1–6, IEEE, 2017.

      4. Al-Nafjan, A., Hosny, M., Al-Wabil, A., Al-Ohali, Y., Classification of human emotions from electroencephalogram (EEG) signal using deep neural networ. Int. J. Adv. Comput. Sci. Appl, 8, 9, 419–425, 2017.

5. Tseng, K.C., Lin, B.S., Han, C.M., Wang, P.S., Emotion recognition of EEG underlying favourite music by support vector machine. 2013 1st International Conference on Orange Technologies (ICOT), pp. 155–158, IEEE, 2013.

      6. Cheng, C., Wei, X., Jian, Z., Emotion recognition algorithm based on convolution neural network. 2017 12th International Conference on Intelligent Systems and Knowledge Engineering (ISKE), IEEE, pp. 1–5, 2017.

      7. Ambler, T., Braeutigam, S., Stins, J., Rose, S., Swithenby, S., Salience and choice: Neural correlates of shopping decisions. Psychol. Marketing, 21, 4, 247–261, 2004.

      8. Khushaba, R.N., Greenacre, L., Kodagoda, S., Louviere, J., Burke, S., Dissanayake, G., Choice modeling and the brain: A study on the Electroencephalogram (EEG) of preferences. Expert Syst. Appl., 39, 16, 12378–12388, 2012.

      9. Vecchiato, G., Kong, W., Giulio Maglione, A., Wei, D., Understanding the impact of TV commercials. IEEE Pulse, 3, 3, 3–65, 2012.

      10.