International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 09 Issue: 06 | June 2022
p-ISSN: 2395-0072
www.irjet.net
A Review on Intelligence Quotient prediction Based on Human Brain Smitha R1, Parvathy G S2 Lecturer, Dept. of Electronics & Communication Engineering, NSS Polytechnic College, Kerala, India Lecturer, Dept. of Computer Engineering, NSS Polytechnic College, Kerala, India ---------------------------------------------------------------------***--------------------------------------------------------------------Abstract –The intelligence of a person is his/her ability to 2. LITERATURE REVIEW learn, understand and solve problems. The intelligence quotient (IQ) is a metric for measuring a person's intelligence. IQ measurements help diagnose various mental disorders, like autism and other conditions. Human intelligence is closely linked to brain function, and so different levels of intelligence are caused by differences in certain brain areas and neural factors. There is little research on detecting someone's IQ using human physiological characteristics rather than skill tests. A review of intelligence quotient prediction based on human brain images and signals is presented in this paper.
A machine learning-based prediction modeling approach was employed to see if intelligent scores could be predicted from spatially highly defined (voxel-wise) patterns of regional gray matter volume [3]. Across functional brain networks, gray matter volume has been found to be an excellent predictor of individual differences in intelligence. A person's IQ score is classified using the transfer learningbased CNN approach of [4] into four classes based on his or her IQ. The MRI (Magnetic Resonance Image) of the brain is used to determine a person's IQ. For the classification task, four pre-trained CNN models were utilized. Out of the four models tested, ResNet-50 had the best accuracy rate with 85.95%.
Key Words:
Intelligence Quotient (IQ), Magnetic Resonance Image (MRI), Electroencephalogram (EEG), Convolutional Neural Network, Support Vector Regression (SVR), Small Visual Geometry Group (SVGG), Visual Geometry Group (VGG), Residual Network (ResNet)
An extended dirty model-based feature selection method has been proposed in [5] to predict IQ scores from brain-MR images. The correlation coefficient of 0.718 and average root mean square error of 8.695, between the estimated and actual IQ values, are obtained when using multi-kernel Support Vector Regression (SVR). SVR with single kernels produced an average correlation coefficient of 0.684 and an average root mean square error of 9.166.
1. INTRODUCTION Although intelligence has been defined in different ways, it encompasses how well we properly plan, reason, wisely resolve issues, learn and draw conclusions, and, in the end, survive in today's world. An IQ test generates a score known as the "Intelligence Quotient" which is used to determine a person's cognitive strength and capabilities [1]. An IQ test determines how well you can perform on one. It tests pattern recognition and data manipulation skills. Practice and persistence can help you improve your score. In an online study on the intelligence quotient (IQ), researchers found the test's scores may not accurately reflect someone's intelligence. They discovered no single test could effectively assess a person's ability to do mental and cognitive activities.
Based on the Brainnetome-Atlas, an evaluation framework including advanced feature selection and regression approaches has been used to predict IQ scores. A Brainnetome-Atlas-based functional connectivity assessment was used to estimate IQ scores in [6]. Fine-grained parcellations of brain networks are provided by Brainnetome Atlas. A predictive framework integrating advanced feature selection and regression approaches have been used to determine continuous IQ scores in a sample of 360 college students. Five regression models were compared to determine which performs best at predicting continuous IQ scores. Moreover, the method for predicting male and female patients is different due to gender differences in the neurobiology of intelligence.
Furthermore, researchers examined the MRI scans of participants' brains and found distinct cognitive abilities were linked to different brain circuits, which confirms that different parts of the brain govern distinct abilities [2]. Because the size, structure, and activity level of different areas of the brain have all been linked to intelligence in humans, detecting the intelligence of an individual from brain images or brain signals will be more accurate. This paper reviews various methods of human intelligence classification and recognition based on the brain. The different methods of Intelligence quotient recognition based on the human brain are explained in Chapter 3. Chapter 4 shows the comparison of the different methods. The conclusion of the study is presented in Chapter 5.
© 2022, IRJET
|
Impact Factor value: 7.529
An effective approach for calculating an individual's IQ using their EEG in resting eye conditions has been presented in [7]. The prediction model is based on sub-band power ratio features from the left hemisphere of the brain and an artificial neural network (ANN). It has been discovered that various groups of intelligence quotient can be categorized based on brainwave sub-band power ratios with 100% training and 88.89% testing accuracy.
|
ISO 9001:2008 Certified Journal
|
Page 1926