Neuro-physiological porn addiction detection using machine learning approach

Abstract

<span>Pornography is a portrayal of sexual subject contents for the exclusive purpose of sexual arousal that can lead to addiction. The availability and easy accessibility of the Internet connectivity have created unprecedented opportunities for sexual education, learning, and growth for adolescences to be in the rise. Hence, the risk of porn addiction developed by teenagers has also increased due to highly prevalent porn consumption. To date, the only available means of detecting porn addiction is through questionnaire. However, while answering the questions, participants may suppress or exaggerate their answers because porn addiction is considered taboo in the community. Hence, the purpose of this project is to develop an engine with multiple classifiers to recognize porn addiction using electroencephalography (EEG) signals and to compare classifiers performance. In the experimental study, the neuro-physiological signals of EEG data were collected previously in Indonesia among students age 9 to 13 years old by researchers from the International Islamic University Malaysia (IIUM). The EEG data were pre-processed, and relevant features are extracted using Mel-Frequency Cepstral Coefficients (MFCC). Then, the features are classified to produce the outputs of valance and arousal. Subsequently, three different classifiers of Multilayer Perceptron (MLP), Naive Bayesian (NB), and Random Forest (RF) are employed to determine whether the participant is a porn addict or otherwise. The experimental results show that the MLP classifier yields slightly better accuracy compared to Naïve Bayes and Random Forest classifiers making the MLP classifier preferable for porn addiction recognition. Although this work is still at infancy stage, it is envisaged for the work to be expanded for comprehensive porn addiction recognition system so that early intervention and appropriate support can be given for the teenagers with pornography addiction problem.</span>

Detecting learning disabilities based on neuro-physiological interface of affect (NPIoA)

“…The frequencies implemented in this research study are delta (1-3Hz), theta (4-8Hz), alpha (9-13Hz), beta (14-25Hz) and also gamma (26Hz-40Hz). The processed EEG data were then used to extract features based on the Mel Frequency Cepstral Coefficients (MFCCs) technique which has been applied in several studies for brain signals analysis [8][9][10]. After features were extracted, at each instance target was labelled based on the emotions data used indicating positive emotion calm with labelled (1, -1) and for negative emotion (fear) with labelled (-1, 1) for valence and arousal (V, A) respectively.…”

“…There are also several proposed machine-learning methods for the EEG sub bands in assisting the diagnosis of learning disabilities based on different focus and analysis based on Discrete Wavelet Transform (DWT), Shannon Entropy Vector and also Artificial Neural Network (ANN) classifier [11]. Since our earlier work using the multi-layer perceptron (MLP) were suggested and has shown to be viable in this paper we will propose the use of MLP as the classification methods to identify the EEG emotions, addiction, behavior and others [8,12,13].…”

An image-based gangrene disease classification

“…In most of the research works [19,20] related to disease classification, image processing was in the crucial part of disease diagnosis process using machine learning algorithms. Image pre-processing is the first step in image processing and pre-processing helps to improve the quality of the image [21] so that the image is suitable to extract the features.…”

Determining Employee Eligibility in Equalizing Staffing Status Using the Naïve Bayes

“…Meanwhile, the level of accuracy (%) in providing predictive conclusions in this study was measured using Equation (2) [18].…”