TraitLWNet: a novel predictor of personality trait by analyzing Persian handwriting based on lightweight deep convolutional neural network

Based on psychologists’ theories, an individual’s handwriting somehow symbolizes a type of personality trait that can be a projection of the person’s innate. A person’s handwriting is the result of an organized system and has scientific bases that make it possible to analyze and specify individuals’...

Full description

Saved in:
Bibliographic Details
Published in:Multimedia tools and applications 2022-03, Vol.81 (8), p.10673-10693
Main Authors: Anari, Maryam Saberi, Rezaee, Khosro, Ahmadi, Ali
Format: Article
Language:English
Subjects:
Accuracy
Algorithms
Artificial neural networks
Computer Communication Networks
Computer Science
Data Structures and Information Theory
Decision making
Handwriting
Lightweight
Multimedia Information Systems
Neural networks
Personality
Personality traits
Psychologists
Special Purpose and Application-Based Systems
Texts
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Based on psychologists’ theories, an individual’s handwriting somehow symbolizes a type of personality trait that can be a projection of the person’s innate. A person’s handwriting is the result of an organized system and has scientific bases that make it possible to analyze and specify individuals’ nature. This paper presents a novel real-time model based on handwriting samples collected from Persian-speaking people, which predicts their personality traits for the first time. Initially, 400 handwriting samples with a repetition of four different texts and psychological questionnaires and three psychologists’ comments have been collected. The pre-processing step is applied to the image samples and the decision-maker model was designed using a lightweight deep convolutional neural network (LWDCNN) structure. The texts were selected based on the psychologists’ guidance. The meaningful relation between the personality trait characters extracted from Persian handwriting and each of the personality traits of the person under-study is matched to a magnificent extent. Finally, the LWDCNN structure is evaluated based on the training samples. The proposed convolutional neural network provides reasonable accuracy for six different and three overlapping personality traits. Despite computational complexity and little time spent by the designed pre-train network to respond, the deep structure’s error level with limited layers is estimated smaller than 10%. The proposed algorithm’s efficiency has been proved by repeating the experiment and assessing measures such as accuracy and mean squared error (MSE).
ISSN:1380-7501
1573-7721
DOI:10.1007/s11042-022-12295-3