Review articleComparative study for 8 computational intelligence algorithms for human identification
Introduction
The identification of a human has become necessary and important in our time. human identification has applications in many various areas where the identity of a user needs to be determined. security systems, or diseases diagnosis e.g. It used some Computational intelligence to apply this [1]. For this Computational intelligence (CI) Is a subfield of artificial intelligence. CI is concerned with the design and development of algorithms and techniques that allow computer to learn [2]. The main aim of CI algorithm is extracting the information from data in automatic approach through statistical and computational approaches. Computational intelligence has become one of the most active researches areas especially in recent years. CI applications in different areas and tasks, such as natural language processing, syntactic pattern recognition, search engines, life sciences, medical diagnosis, bioinformatics, DNA, robot, computer vision.
Researchers used a number of algorithms for human identification such as neural networks, fuzzy logic, decision tree, … etc. [3]. Usually computational intelligence are a collection of nature inspired computational approaches to address complex real world problem [4]. Biometric uses these algorithm for this. As the biometric passes through several stages in order to be able to determine the identity of a human, shown Fig. 1. These stages are:
- i.
Capturing: using sensor or scraping a website and extract data-set
- ii.
Enrollment: takes the captured information and create a unique the feature for subject
- iii.
Comparison: the biometric system measures the likeness of the unique feature collected at the enrollment stage to the data collected at the current authentication stage
- iv.
Decision: in this stage, the system decides whether or not to give the subject access.
Several algorithms are used to define human identity such as, artificial neural networks, fuzzy logic, genetic algorithm, … etc. Therefore we conduct comparative study to evaluate the performance of these methods.
The challenges associated with human identification can be attributed to the following factors.
- i.
Imaging conditions: Refer to when formed images, there are some factors that affect the quality of the image.
- ii.
Universality: people without fingers or with injured, eyes, these cases must be handled
- iii.
Collect: obtain the feature should be easy and simple
- iv.
Age: when Image taken after one or two year’s gap may not match with the image in data-base.
With regard about choosing in such research. It is necessary to search for algorithms that give us high accuracy results and also the cost that became very necessary in application at a time when the world became focused on material issues. There are technologies that have been developed in machine learning. These techniques are used for verification/authentication or diagnosis of diseases. Table 1. Explain illustrates the differences between subject identifiers [5].
Many studies trying to provide a good features and identification systems such as R. Chatterjee, et al. (2019) [5] Proposed a novel features selection approach for Motor Imagery EEG waves Classification when used BCI Competition as data-set. Various fuzzy & non-fuzzy dissimilarity measures was used to determine feature. More than 70% was obtained from the first 50% and 37% of selected features for 24 and 12 features data-sets respectively. A. Y. Ng, and M. I. Jordan (2002) [6] analyzed logistic regression and Discriminant Analysis for input(especially continuous inputs). Fifteen of the data-set were applied, eight with continuous input, seven with discrete input from, the UCI Machine Learning repository. Found Fuzzy Discernibility Matrix algorithm sometimes with Large dimension may not necessarily give us relevant information of a problem, but computing a minimal relevant dataset cost high. Another study by C.-L. Lim, et al. (2011) [7] applied fast computation approach for exact Zernike moments by using digital filter. Grayscale image as data-set. The computation time when used digital filter needs only half the time required to do the computation using Hosny’s computation technique. Eager Decision Tree was proposed by S. S. Gavankar and S. D. Sawarkar (2017) [8]. The best approach to address the messiness in test data is Lazy Decision Tree, where attribute with missing value are avoided by constructing the tree at testing time with known attribute. Also, Genetic algorithm was applied by Senthilkumaran, N (2012) [9] when Dental X-ray edge detection to produce the edges of the image. High performance was present when used this algorithm in his/her research.
This paper aims to present a comprehensive survey about the application of CI based methods in person identification/verification. Eexplanation some classification techniques in machine learning and algorithms being used for each technique. Also analyzing large amounts of algorithms to predict which one is better for human identification according to execution time, secure, storage capacity of the algorithm application, and the effort to implement the algorithm. In addition, the advantages and disadvantages of each approach and results obtained by the suggestions of researchers and dataset used in these techniques. During the previous years, the field of human identification has become one of the, broadest areas of research. Especially when applied algorithm for human identification.
The paper is structured as follows: Section 2: Computational intelligence techniques for human identification such ask-Nearest Neighbor(K-NN), Artificial Neural Networks (ANNs), Support vector machines (SVMs), Fuzzy Discernibility Matrix (FDM), Naïve Bayes (NB), k-means, Decision Trees (DTs), and Genetic algorithms (GAs). Section 3: implementation of previous methods. Section 3: Rsult and Discussion and Last section conclusion
Section snippets
Computational intelligent techniques for human identification
In this section, several methods used human identification based on biometrics are vividly discussed.
Result and discussion
This section presents the result and explanation of our analysis in this study. There are some algorithms that are distinct from other. It was found that SVM algorithm is more accuracy than KNN classifier. While KNN classifier has a faster execution time than SVM according to R. Chatterjee, et al. (2019) [5]. AdaBoost and SVMs have more efficiency than K-NN classifier. In addition the KNN approach is more dominant than SVMs. But Naïve Bayes model is outperforms the Logistic regression P.
Conclusion
human identification is one of the challenging aspect such as image analysis and computer vision. The focus towards the human identification has been increased in the last years due to its enormous applications in various domains. For this reason it was used Computational intelligent to facilitate the task of identifying the human. So that the CI work as an approach on noise-affected or incomplete data to obtain robust and approximate solutions for one-human identification. demonstrated some.
CRediT authorship contribution statement
Shaymaa Adnan Abdulrahman: Writing - original draft, Writing - review & editing. Wael Khalifa: Supervision. Mohamed Roushdy: Supervision, Investigation. Abdel-Badeeh M. Salem: Supervision, Methodology, Project administration.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References (80)
- et al.
Solving vehicle routing problem by using improved K-nearest neighbor algorithm for best solution
J. Comput. Sci.
(2017) - et al.
Fast computation of exact zernike moments using cascaded digital filters
Inform. Sci.
(2011) - et al.
Landslide spatial modeling: Introducing new ensembles of ANN, MaxEnt, and SVM machine learning techniques
Geoderma
(2017) An extension of the naive bayesian classifer
Inform. Sci.
(2006)- et al.
Examining multiple feature evaluation and classification methods for improving the diagnosis of Parkinson’s disease
Cogn. Syst. Res.
(2019) - et al.
Comparison of a logistic regression and naive bayes classifer in landslide susceptibility assessments: The inuence of models complexity and training dataset size
Catena
(2016) - et al.
Neural network and multi-fractal dimension features for breast cancer classification from ultrasound images
Comput. Electr. Eng.
(2018) - et al.
Artificial neural network modelling of amido black dye sorption on iron composite nano material: kinetics and thermodynamics studies
J. Molecular Liquids
(2018) - et al.
Predicting future hourly residential electrical consumption: A machine learning case study
Energy Build.
(2012) - et al.
A deep convolutional neural network model to classify heartbeats
Comput. Biol. Med.
(2017)
Fuzzy rule based decision trees
J. Pattern Recognit. Elsevier
Solving vehicle routing problem by using improved genetic algorithm for optimal solution
J. Comput. Sci.
Genetic case-based reasoning for improved mobile phone faults diagnosis
Comput. Electr. Eng.
Fuzzy expert system for predicting pathological stage of prostate cancer
Expert Syst. Appl.
Comparing the analytical performances of micro-nir and fafnir spectrometers in the evaluation of acerola fruit quality, using PLS and SVM regression algorithms
Talanta
Electroencephalogram signal classification based on shearlet and contour let transforms
Expert Syst. Appl.
A real time computer aided object detection of nasopharyngeal carcinoma using genetic algorithm and artificial neural network based on haar feature fear
Future Gener. Comput. Syst.
Deep convolutional neural network for the automated detection and diagnosis of seizure using eeg signals
Comput. Biol. Med.
Vocal acoustic analysis-classification of dysphonic voices with artificial neural networks
Procedia Comput. Sci.
Real time wave forecasting using neural networks
Ocean Eng.
Novel naphthylstyryl-pyridinium potentiometric dyes offer advantages for neural network analysis
J. Neurosci. Methods
A novel cancelable iris recognition system based on feature learning techniques
Inform. Sci.
Predicting sex as a soft-biometrics from device interaction swipe gestures
Pattern Recognit. Lett.
Biometric cattle identification approach based on weber’s local descriptor and adaboost classifier
Comput. Electron. Agric.
Fuzzy rule-based decision trees
Pattern Recognit.
Application of k-means and genetic algorithms for dimension reduction by integrating SVM for diabetes diagnosis
Procedia Comput. Sci.
Authentication of smartphone users using behavioral biometrics
IEEE Commun. Surv. Tutor.
Multitask learning without label correspondences
Eeg-based biometric authentication modelling using incremental fuzzy-rough nearest neighbour technique
IET Biom.
Leave-one-out authentication of persons using 40 hz eeg oscillations
On discriminative vs. generative classifiers: A comparison of logistic regression and naive bayes
Eager decision tree
Genetic algorithm approach to edge detection for dental x-ray image segmentation
Int. J. Adv. Res. Comput. Sci. Electron. Eng. (IJARCSEE)
Eeg-based emotion recognition via fast and robust feature smoothing
Human identification system: Areview
Int. J. Comput. Bus. Res. (IJCBR)
Transfer learning and supervised classifier-based prediction model for breast cancer
Discrimination of mine seismic events and blasts using the _sher classifier, naïve bayesian classifier and logistic regression
Rock Mech. Rock Eng.
Comparative study of k-nn naive bayes and decision tree classification techniques
Int. J. Sci. Res. (IJSR)
Cited by (22)
GeneNet: Transfer learning-based hybrid African buffalo optimization with genetic algorithm for gene expression based cancer classification
2023, e-Prime - Advances in Electrical Engineering, Electronics and EnergyAn optimization for adaptive multi-filter estimation in medical images and EEG based signal denoising
2023, Biomedical Signal Processing and ControlCitation Excerpt :In [21–23], authors have presented a comprehensive survey for the application of machine learning techniques in medical data processing which involves image-signal denoising, classification, regression, and clustering. Along with classical machine learning methods, some soft computing techniques are widely used in the medical domain since highly convergent bio-inspired methods have been proposed in recent years [24]. Disease images also require some EEG-based neural signals for efficient diagnosis.
Importance analysis of decision making factors based on fuzzy decision trees
2023, Applied Soft ComputingA comprehensive survey on the biometric systems based on physiological and behavioural characteristics
2023, Materials Today: ProceedingsClassification of breast cancer using microarray gene expression data: A survey
2021, Journal of Biomedical InformaticsCitation Excerpt :However, the simple assumption of NB, it is fast and effective in real problems. Bayesian belief networks are used to deal with the features’ dependency [20]. Support Vector Machine (SVM) is commonly used for classifying gene expession data due to the sparseness of solution sparseness of solution and it’s ability to handle large feature space [22].
- 1
PhD Student at Ain Shams University, Egypt.