Machine Learning Methods for Fear Classification Based on Physiological Features

Abstract

This paper focuses on the binary classification of the emotion of fear, based on the physiological data and subjective responses stored in the DEAP dataset. We performed a mapping between the discrete and dimensional emotional information considering the participants’ ratings and extracted a substantial set of 40 types of features from the physiological data, which represented the input to various machine learning algorithms—Decision Trees, k-Nearest Neighbors, Support Vector Machine and artificial networks—accompanied by dimensionality reduction, feature selection and the tuning of the most relevant hyperparameters, boosting classification accuracy. The methodology we approached included tackling different situations, such as resolving the problem of having an imbalanced dataset through data augmentation, reducing overfitting, computing various metrics in order to obtain the most reliable classification scores and applying the Local Interpretable Model-Agnostic Explanations method for interpretation and for explaining predictions in a human-understandable manner. The results show that fear can be predicted very well (accuracies ranging from 91.7% using Gradient Boosting Trees to 93.5% using dimensionality reduction and Support Vector Machine) by extracting the most relevant features from the physiological data and by searching for the best parameters which maximize the machine learning algorithms’ classification scores.

Multi-Input CNN-LSTM deep learning model for fear level classification based on EEG and peripheral physiological signals

“…The data of both studies is the EEG signal converted to the alpha, beta, and theta frequencies. Petrescu et al (2021) reported a binary classification of fear with an average accuracy of 92.40% during 10 repeated 7:3 holdouts by reducing the dimensionality with Principal Component Analysis of the GSR and Plethysmographs of the DEAP dataset for inputting to SVM. With the MANHOB dataset ( Soleymani et al, 2012 ), Miranda et al (2021) performed binary classification of fear with an accuracy of 96.33% with five-fold cross-validation using a classifier composed of SVM and kNN, whose inputs were temporal, frequency, and non-linear based features derived from ECG, Skin Temperature, and GSR signals of only female participants of the dataset.…”

“…Although the studies of emotion-level recognition are everything described above as far as we investigated, few of them use the Deep Learning model approach based on Deep Neural Networks. In many cases, the Machine Learning model approach is more accurate than the Deep Learning model approach for emotion-level recognition ( Bălan et al, 2019 , 2020 ; Petrescu et al, 2021 ). However, the Machine Learning model approach is required feature extraction and feature selection to improve accuracy.…”

Quantitative identification of ventral/dorsal nerves through intraoperative neurophysiological monitoring by supervised machine learning

“…All models were used with the percentage of training of 10-fold cross-validation for the classification, with metrics including accuracy, precision, recall (sensitivity), F1 score, and specificity correspondingly calculated. A detailed definition of these metrics can be found in the paper published by Petrescu ( 17 ).…”

Analysis of Marine-Pilot Biometric Data Recordings during Port-Approach Using a Full-Mission Simulator

“…Recognizing and prioritizing the important navigational information in the officers’ working memory, situation awareness, and making appropriate decisions may indirectly affect the participants’ body response, which is manifested in excessive heart rate (HR), blood volume pulse (BVP), electrodermal activity (EDA), and pupil diameter. Related work found behavioural performances to be possible causes of such effects, which will be the scope of our future research [ 17 , 18 , 19 ].…”