N. Taghvaei

Bio: N. Taghvaei is an academic researcher from Qazvin Islamic Azad University. The author has contributed to research in topics: Feature (machine learning) & Categorization. The author has an hindex of 1, co-authored 1 publications receiving 1 citations.

A Hybrid Framework for Personality Prediction based on Fuzzy Neural Networks and Deep Neural Networks

Abstract: In general, humans are very complex organisms, and therefore, research into their various dimensions and aspects, including personality, has become an attractive subject of research. With the advent of technology, the emergence of a new kind of communication in the context of social networks has also given a new form of social communication to humans, and the recognition and categorization of people in this new space have become a hot topic of research that has been challenged by many researchers. In this paper, considering the Big Five personality characteristics of individuals, first, categorization of related work is proposed, and then a hybrid framework based on Fuzzy Neural Networks (FNN), along with, Deep Neural Networks (DNN) has been proposed that improves the accuracy of personality recognition by combining different FNN-classifiers with DNN-classifier in a proposed two-stage decision fusion scheme. Finally, a simulation of the proposed approach is carried out. The proposed approach is using the structural features of Social Networks Analysis (SNA), along with a linguistic analysis (LA) feature extracted from the description of the activities of individuals and comparison with the previous similar researches. The results, well-illustrated the performance improvement of the proposed framework up to 83.2 % of average accuracy on myPersonality dataset.

DENOVA: Predicting Five-Factor Model using Deep Learning based on ANOVA

Abstract: Determining the personality dimensions of individuals is very important in psychological research. The most well-known example of personality dimensions is the Five-Factor Model (FFM). There are two approaches 1- Manual and 2- Automatic for determining the personality dimensions. In a manual approach, Psychologists discover these dimensions through personality questionnaires. As an automatic way, varied personal input types (textual/image/video) of people are gathered and analyzed for this purpose. In this paper, we proposed a method called DENOVA (DEep learning based on the ANOVA), which predicts FFM using deep learning based on the Analysis of variance (ANOVA) of words. For this purpose, DENOVA first applies ANOVA to select the most informative terms. Then, DENOVA employs Word2Vec to extract document embeddings. Finally, DENOVA uses Support Vector Machine (SVM), Logistic Regression, XGBoost, and Multilayer perceptron (MLP) as classifiers to predict FFM. The experimental results show that DENOVA outperforms on average, 6.91%, the state-of-the-art methods in predicting FFM with respect to accuracy.

Graphology analysis for detecting hexaco personality and character through handwriting images by using convolutional neural networks and particle swarm optimization methods

Abstract: Graphology or handwriting analysis can be used to infer the traits of the writers by examining each stroke, space, pressure, and pattern of the handwriting. In this study, we infer a six-dimensional model of human personality (HEXACO) using a Convolutional Neural Network supported by Particle Swarm Optimization. These personalities include Honesty-Humility, Emotionality, eXtraversion, Agreeableness (versus Anger), Conscientiousness, and Openness to Experience. A digital handwriting sample data of 293 different individuals associated with 36 types of personalities were collected and derived from the HEXACO space. A convolutional neural network model called GraphoNet is built and optimized using Particle Swarm Optimization (PSO). The PSO is used to optimize epoch, minibatch, and droupout parameters on the GraphoNet. Although predicting 32 personalities is quite challenging, the GraphoNet predicts personalities with 71.88% accuracy using epoch 100, minibatch 30 and dropout 52% while standard AlexNet only achieves 25%. Moreover, GraphoNet can work with lower resolution (32 x 32 pixels) compared to standard AlexNet (227 x 227 pixels).