Volume 9, Issue 8, August – 2024 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24AUG265

Building Face Ageing Model Using Face Synthesis

Shraddha Mishra Manvi Chahar Shivani Jaswal
Department of CSE Department of CSE Department of CSE
Apex Institute of Technology Apex Institute of Technology Apex Institute of Technology
Chandigarh University Punjab, India Chandigarh University Punjab, India Chandigarh University Punjab, India

Abstract:- Advancements in face synthesis technology Both the networks grind their skills through an iterative
have enabled innovative methods for modeling facial adversarial procedure, giving results in the creation of
aging. This research paper focuses primarily on creating lifelike images. After merging computer vision, machine
a robust face aging model using deep learning and learning, and image processing, the main purpose of GANs
Generative Adversarial Networks (GANs), trained on a for addressing aging synthesis tells how to transform the
diverse dataset of facial images. The proposed approach several fields and fundamentally transform the understanding
captures both global features and local textures to of aging dynamics.
produce realistic age-progressed images while preserving
the subject's identity. This paper also examines face This article addresses unsolved issues by upgrading the
synthesis techniques, with specific emphasis for the generation procedure as a discriminator's performance and
various practical usage of GANs. The key objective of our providing a thorough overview of the most recent
project is to upgrade both the discriminator and the developments in face aging synthesis using GANs. In this we
generator parts of GANs to generate more realistic, age- researched the underneath concepts behind GANs and how
progressed face images. We evaluated the model using they are used for generating realistic facial images. We then
quantitative metrics and qualitative assessments, examined a wide range of methods which focused at
demonstrating its effectiveness. Additionally, we address negotiating with certain situations which are associated with
ethical considerations, proposing guidelines for face aging, such as time progression reliability, authenticity,
responsible use. Our study offers a novel framework with and personality preservation.
significant applications in security, forensics, and
entertainment, and suggests future research directions to Further , we have explained how evaluation demands
improve accuracy and ethical standards. and standard datasets portray a vital role while assessing the
effectiveness of face aging models and clarify the subtle
Keywords:- Generative Adversarial Networks (GANs), aspects of sensation stature, age advancement commitment,
Convolutional Neural Networks (CNNs), Generator, and identity preservation. We have also looked at the
Discriminator, IP-CGAN. revolutionary potential of GAN-based face aging synthesis in
a variety of operations, from digital amusement platforms to
I. INTRODUCTION fingerprint verification systems, furthermore in the technical
components.
To try to smooth the puzzles of time's transit engraved
on human figures, the marriage of artificial intelligence and However regardless of the clear future, there are the
photographic synthesis has appeared as a transformative force. meaningful ethical questions and society implications to
This aging phenomenon of face, has basically natural, ponder. The putting together of intentionally aged images
enchanted researchers and expounders across disciplines, opens up questions about permission, privacy, and the
from forensic science to entertainment, outstanding its prospective abuse of synthetic data. Therefore , we enhance
profound implications and diverse applications. The face the value of responsible invention and ethical governance in
aging field of face synthesis has adapted an experienced a exploiting the energize of GANs for face aging synthesis.
prototype change with the development of deep learning
approaches, primarily Generative Adversarial Networks Finally, the combination of GANs with face
(GANs), which have generated in previously unattainable development synthesis signals the start of a new era of the
levels of naturalism and consistency in age progression. technological advancement by giving never-before-seen
circumstances to solve the problems of time as well as
Previously, age cycle methods mostly depended on convert how we relate to aging. While we control the future
human adaptations or fundamental programming that couldn't ethical and societal issues that we will work is to fully
adequately behold the exclusiveness of facial aging. However, discover the possible of this creative upgraded by serious
the introduction of GANs signaled a revolution in the field of research and worthy implementation.
generating various models. The primary concept behind how
GANs function is that ,how a discriminator network tries to
differentiate between the actual and artificial samples, while a
generator network make efforts achieving aim to create
realistic data samples.

IJISRT24AUG265 www.ijisrt.com 694

Volume 9, Issue 8, August – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24AUG265

A. Problem Statement (Super-Resolution Generative Adversarial Network), the

Traditional approaches of face synthesis often have generator network aspires to depict images of low resolution
failed to produce the realistic aged images with the into the images with the high-resolutions, whereas the
preservation of the identity of the input images. They also discriminator learns to differentiate between real images and
lack the ability to balance the identity preservation and age the images generated by the generator.
progression which result in generating the images that lack
fidelity and are inconsistent. Therefore, it was needed to We observed that in Kaneko et al [4], it provides a
develop such approach that can produce high fidelity images thorough examination of Generative Adversarial Networks
and also maintaining the characteristics of the input images. (GANs), while illuminating the diverse applications and
Hence, we have implemented Identity Preserved CGAN principles across multiple domains. This paper examines the
approach that is capable of fulfilling the objective. This paper detailed underlying techniques of GANs, where a generator
provide a robust solution for producing the aged faces while network and a discriminator network engage in competition,
preserving the identity and improving the generator and with the generator striving to produce outputs of heightened
discrimination performance. realism. Furthermore, this paper also highlighted the
applications of GANs, spanning image generation, data
B. Objective augmentation, image-to-image translation, and so forth. Y.
To develop a face aging application using face synthesis Singh et al [5] described various activation functions for
techniques which helps to generate aged face images using the classification problems and suggested some more suitable
young inputs and also increasing the performance of the activation function for specification problems.
generator and discriminator. The objective includes building a
robust model based on Generative Adversarial Networks in In paper by Sharma et al [6] a fresh approach to face
order to synthesize a realistic image. Moreover, the paper also aging synthesis has been presented. They focused on feature
seeks at enhancing the generator and discriminator’s fusion techniques to enhance the realism and accuracy of age-
performance. To increase the potency of the model both progressed facial images. Their approach offers a more
qualitative and quantitative analysis will be conducted. extensive and refined depiction of facial aging dynamics, by
integrating feature-level information from diverse sources,
II. RELATED WORK including facial landmarks, texture patterns, and framework
characteristics. They utilized Generative Adversarial
Some various approaches for face aging are outlined in Networks (GANs) for the advancement of cutting-edge face
Pranoto, H et al [1]. This paper described Conventional age progression. The study deals with the growing
Approach, Deep Generative Model for Face Aging Approach significance of facial age progression techniques in many
and Generative Adversarial Networks Approach. applications, such as law enforcement, entertainment, and
Conventional approach includes Model-Based Approach, forensics, highlighting the need for realistic and high-quality
Prototype Approach and Reconstruction-Base Approach. transformations. This research adds to the ongoing
Deep Generative Model for Face Aging Approach includes advancement of facial image processing techniques, offering
Temporal Restricted Boltzmann Machine-Based Model promising paths for applications requiring age progression
(TRBM), Recurrent Neural Network-Based Model and and super-resolution capabilities.
Generative Adversarial Networks Approach. Generative
Adversarial Networks Approach includes Translation-Based Z. Zhang et al [7] introduced a sophisticated approach to
Approach, Triple-GAN, Child Face Age-Progression VIA face aging utilizing an innovative structure termed A3GAN.
Deep Feature Aging, Sequence Based Approach, Subject The proposed A3GAN model adequately captures and
Dependent Deep Aging Path (SDAP) Model and Conditional- preserves distinctive facial features while synthesizing
Based Approach. realistic aged faces by integrating a feature-aware attentive
mechanism within a Generative Adversarial Network (GAN)
The use of Conditional GAN (CGAN) can be seen in architecture. This feature-awareness enables the network to
Antipov, G et al [2], where extra details such as age labels are concentrate on particular facial attributes, enhancing the
provided to both the generator and discriminator. This accuracy of age progression results.
addition allows for more accurate control over the process of
aging, ensuring that the images generated display age- X. Tang et al [8] demonstrated an alternative framework
appropriate characteristics while maintaining identity features. for training generative models, Generative Adversarial Nets.
The conditional version of generative adversarial nets are
A combined approach for face age progression using proposed, which can be developed by feeding the data. Their
Attention-GAN and SRGAN is presented in Neha Sharma et model consisted of two adversarial models: a generative
al [3]. Attention-GANs utilize attention mechanisms to model which holds the data distribution, and a discriminative
selectively attend to facial regions that are more prone to age- model that evaluates the probability that a sample came from
related changes, such as wrinkles, fine lines, and changes in the training data. They also illustrated how this model can be
skin texture. By emphasizing these regions during the used for learning a multi-modal model, and provide early
generation process, Attention-GANs can produce age demonstrations of an application to image categorization.
progressed images that exhibit more realistic aging effects
while preserving the identity of the individual. In SRGAN

IJISRT24AUG265 www.ijisrt.com 695

Volume 9, Issue 8, August – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24AUG265

Conditional GANs (CGANs) take advance information training processes were carried out using TensorFlow which
for image generation and construct the generated images with facilitates efficient computation and scalability.
suitable target age. Paper [9] presented an Identity-Preserved
Conditional Generative Adversarial Networks (IPCGANs) for For evaluating the performance of the trained GAN,
face aging, motivated by the accomplishments of CGANs. quantitative and qualitative metrics were done. Quantitatively,
the generator and discriminator both are analysed on the basis
Generative Adversarial Networks (GANs) have emerged of loss weights which carry out the training process. Different
as an influential framework for face aging models because of metrics are included such as generative adversarial loss,
their potential to produce a realistic as well as superior top feature matching loss, and age classification accuracy to
notch images. GANs consist of two main components: a assess the quality and variety of the generated aged faces.
generator and a discriminator. The generator which is Qualitatively, visual inspections and human evaluation studies
frequently depicted as deep neural network takes input of were conducted so as to compare the generated images with
random noises and converts them into texts or images while ground truth aged faces from the dataset collected from
the discriminator distinguishes between real/actual and Kaggle, ensuring the visual realism and perceptual quality of
generated images. the result.

III. METHODOLOGY The proposed methodology was implemented using

Python programming language, through TensorFlow for
In recent, face ageing application has gained interests model training and evaluation. Extensive experimentation was
because of its wide range applications in fields of conducted for exploring several hyperparameters, network
entertainment, forensics etc.. However, the obstacle lie in the architectures, and training strategies. For feature extraction
generation of the aged faces from the young face inputs. and representation learning, CNNs were employed, while
Therefore, Generative Adversarial Networks, a face synthesis machine learning techniques were utilized for data analysis
techniques provide a promising solution in this direction by and model optimization.
producing a high quality images. It consists of 2 neural
network which helps in one aims at generating the realistic Python programming which is a versatile programming
images whereas other differentiate between the real and the language, is used to carry out the implementation process
fake images. because of its readability and rich libraries. CNN plays a vital
role the Identity preserved CGANs framework for both the
A methodological approach to building face aging generator and the discriminator. CNNs abilities helps to
applications using face synthesis is presented in this paper identify and extract common image patterns vital for
having main and specific focus on improving the performance generating the images along with identity preservation. Image
of the discriminator and generator components of the processing techniques help in pre-processing the training data
adversarial network. The primary objective was to develop a and face detection, alignment etc. are done to normalize the
robust and efficient system which is capable of generating input images. An open-source ML platform i.e. TensorFlow is
realistic aged faces images from young face inputs. This is used to train and deploy the network because of its robustness
achieved through machine learning, Python programming in handling the large datasets.
language, Convolutional Neural Networks (CNNs), and image
processing techniques. The output of the experiments demonstrated the
potencial of the proposed methodology in improving the
The qualitative data that we utilized was collected from performance of the discriminator and generator in the face
Kaggle, a widely-used platform for sharing datasets and aging applications. We have presented the quantitative metrics
conducting data-driven research. The dataset collected had and qualitative assessments, showing the quality and realism
been pre-processed so as to ensure the consistency and quality of the generated aged faces. We have compared the
in the inputted data. Facial alignment to normalize face performance and discussed the implications of our findings of
orientations, resize images to a standard resolution had been our approach with existing techniques, highlighting the
performed and then pixel values are normalized to a common advantages or limitations.
scale. Moreover, we had employed data augmentation
techniques I.e. rotation, flipping, and scaling so as to increase An important factor of building face aging applications
variability of the dataset and also improving the robustness of using face synthesis was the selection of an appropriate GAN
the trained model. architecture and hence evaluated several architectures
including IPCGAN, DCGAN, StyleGAN, and Pro-GAN,
An iterative training process had been adopted for the considering the factors such as computational efficiency and
GAN, where both the generator and discriminator networks image quality. Based on the evaluation, we chose a GAN
were trained simultaneously. During training, appropriate loss architecture that had strike a balance between these factors
functions are defined including adversarial loss, feature and was well-suited for the tasks of face synthesis. The
perceptual loss, and matching loss for guiding the generator and discriminator networks were designed using
optimization of the discriminator and generator. Transfer CNN architectures which is optimized for image generation
learning techniques might also be utilized, initializing the and discrimination. TensorFlow was also employed because
GAN with pre-trained models to expedite convergence. The

IJISRT24AUG265 www.ijisrt.com 696

Volume 9, Issue 8, August – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24AUG265

of its efficiency and flexibility in implementing deep learning A. Network Architecture

models. In this paper, we have implemented the IP-CGAN i.e.
Identity Preserved Conditional GAN which is used because of
In the paper, we have shown a methodological approach its dual objective. Firstly, it generates realistic aged images
for building face aging applications using the approach GANs from the young face inputs and also preserve the identity of
architecture such as IPCGAN and CGAN which produce the the inputted image.
aged face images from the young inputs provided and also
having a special focus on improving the performance of the
discriminator and generator using GANs.

Fig. 1. Network Architecture

In IP-GAN generator comprises of several convolutional IV. IMPLEMENTATION

layers, un-sampling layers and residual blocks. The layers are
designed to transform the input face images into the aged face A. Experimental Setup
images and also preserving the identity features. Figure 1 For the paper, we have explored face aging effects using
shows the network architecture of the model. It leverages Identity-Preserved Conditional Generative Adversarial
various activation functions and regularization techniques in Networks as presented in paper[10]. We have setup the
order to make smooth and natural age progression. On the experimental environment with the different libraries and
other hand, discriminator network in IP-CGAN differentiate dependencies. TensorFlow-gpu of version 1.4.1 is utilized for
between real and fake images while at the same time preserve the GPU resources. SciPy, OpenCV-python and NumPy are
the identity information. Like generator, it also employs a the core dependencies used for addressing the scientific
series of convolutional layer, down sampling layers, and fully computing and image processing needs. CUDA toolkit is also
connected layers in order to effectively differentiate between installed in order to ensure the compatibility and optimal
real and synthesized images. performance with Tensor Flow GPU.

The research focused on contributing to advancing the B. Dataset

fields of face aging technology, offering a robust and efficient In this paper we have chosen Cross Age Celebrity
solution so as to generate a realistic aged faces from young Dataset from Kaggle for the purpose of training and
face inputs. We anticipated that our proposed methodology evaluation. The dataset holds more than 1 lakh images of
would inspire further research and development in the approximately 2000 celebrities with age between 15 to 65. All
following area and opening new avenues for applications in the images in the dataset have large variation in pose,
various sectors. illumination, expressions and their styles. The dataset is
partitioned into two parts: training and test data where training
data is categorized into groups according to ages.

IJISRT24AUG265 www.ijisrt.com 697

Volume 9, Issue 8, August – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24AUG265

C. Model Evaluation feature matching loss ensures that the images generated
We have implemented the IPCGAN i.e. Identity matched the real images in order to preserve the identity of
Preserved Conditional Generative Adversarial Network the different age groups. These loss functions are important to
approach in our paper. In the paper, we have first compared maintain the balance between the uniqueness of the identity
the different GANs methods with our method which are all characteristics and age progression and thereby increasing the
build on conditional GANs and are quite similar to our performance of the network.
method. We have used TensorFlow- GPU, scipy and numpy
libraries for the implementation of our method. We started E. Improvement
training the data from the scratch where we first downloaded To enhance the performance of the generator and the
the pre-trained model, including AlexNet and an age discriminator within IP-CGANs, AlexNet, a pre trained model
classification model where age classifier is finetuned on and age classification models are used for the purpose of
celebrity training dataset with 200000 steps and the whole quick convergence and better generalization through the
training process complete in about 5 lakh steps. Additionally, transfer learning. Hyperparameter tuning such as adversarial
we employed specific hyperparameters such as age loss loss and the feature matching loss helps ensures a balance
weight, feature loss weight to balance the identity between the networks. Feature extraction layer such as conv5
preservation and the aging effect. We have chosen conv5 ensures the identity preservation along with the introduction
feature layer for the feature extraction so as to optimize face of the age-specific features.
ageing translation.
V. RESULT AND DISCUSSION
D. Loss Function
We have utilized the loss functions for the training of In our paper, we have compared the effectiveness of the
generator and discriminator of the IP-CGANs. The adversarial Identity Preserved Conditional Generative Adversarial
loss function encourages the generator for producing more Networks with many other approaches.
realistic images in order to fool the discriminator and the

Fig. 2. Some Results of IPCGANs

Figure 2 depicts the aging effects of the model. The preserved the unique features of the input images and yield
results showed the higher fidelity image while maintaining the the age-related characteristics.
uniqueness of the identity feature and also achieved a natural
ageing effect. IPCGANs achieves an exceptional accuracy of Quantitative evaluation shows that the approached used
approx. 96 % for the face verification and better image quality in the paper has been able to perform better in terms of
by 71% as compared with the other methods. The result preserving the uniqueness of the identity of all the age groups.
produced with our approach shows it has excellently Through the model, it can be seen that the mean squared error

IJISRT24AUG265 www.ijisrt.com 698

Volume 9, Issue 8, August – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24AUG265

of the input and the generated image is also reduced with the [3]. Sharma, Neha, Reecha Sharma, and Neeru Jindal.
IP-CGANs method. The mean square error score of approx. "Prediction of Face Age Progression with Generative
0.012 and the structural similarity index of 0.85 shows high Adversarial Networks." Multimedia Tools and
fidelity and similarity with the target. This shows the ability Applications 80.25 (2021): 33911-33935.
of the model in preserving the integrity of the image identity [4]. Kaneko, Takuhiro. "Generative Adversarial Networks:
throughout the ageing process. Quantitative assessment also Foundations and Applications." Acoustical Science and
shows that proposed approach has demonstrated higher Technology 39.3 (2018): 189-197.
accuracy in comparison with the baseline. [5]. Y. Singh, M. Saini and Savita, "Impact and
Performance Analysis of Various Activation Functions
Qualitative analysis shows that while doing the visual for Classification Problems," 2023 IEEE International
examination of the result, close resemblance of the generated Conference on Contemporary Computing and
aged faces found with the targeted aged faces which shows Communications (InC4), Bangalore, India, 2023, pp. 1-
the preservation of the identity of the input face. IP-CGANs 7, doi: 10.1109/InC457730.2023.10263129.
has various applications in multiple domains such as digital [6]. Sharma, Neha, Reecha Sharma, and Neeru Jindal. "An
entertainment, security, and helping find missing persons. Improved Technique for Face Age Progression and
This identity-preserved aging model holds promise for Enhanced Superresolution with Generative Adversarial
improving the precision of age-progressed images in these Networks." Wireless Personal Communications 114
highly sensitive domains. It is also important to acknowledge (2020): 2215-2233.
the constraints in our study. First of all our results depends [7]. Z. Zhang, Y. Song, and H. Qi, “Age
upon the quality and diversity of the training dataset utilized. Progression/Regression By Conditional Adversarial
Moreover, the computational resources, necessary for the Autoencoder,” In Proc. IEEE Conf. Comput. Vis.
training and inference, might bring various restrictions and Pattern Recognit. (CVPR), Jul. 2017, Pp. 4352–4360.
demerits to the feasibility of our approach when used with real [8]. X. Tang, Z. Wang, W. Luo, and S. Gao, ‘‘Face Aging
time scenarios. with Identity-Preserved Conditional Generative
Adversarial Networks,’’ In Proc. IEEE/CVF Conf.
VI. CONCLUSION Comput. Vis. Pattern Recognit., Jun. 2018, Vol. 9, No.
1, Pp. 7939–7947, Doi: 10.1109/CVPR.2018.00828.
This paper has proposed improvement to both the [9]. P. D. S. Prasad, R. Tiwari, M. L. Saini and Savita,
network components of the generative adversarial network i.e. "Digital Image Enhancement using Conventional
generator and discriminator. Through the current development Neural Network," 2023 2nd International Conference
in this field, a comprehensive overview has been offered in for Innovation in Technology (INOCON), Bangalore,
the face ageing using GANs approach. IP-CGAN, a GANs India, 2023, pp. 1-5, doi:
methodology is used in the paper with its dual objective of 10.1109/INOCON57975.2023.10100995.
both enhancing the performance and the fidelity of the [10]. M. Sohail, M. Lal Saini, V. P. Singh, S. Dhir and V.
generated aged images from the young inputs along with the Patel, "A Comparative Study of Machine Learning and
identity preservation makes it a valuable tool for the accurate Deep Learning Algorithm for Handwritten Digit
representation of the implemented approach of age Recognition," 2023 6th International Conference on
progression. Moreover, the paper stresses on the necessity of Contemporary Computing and Informatics (IC3I),
ethical scrutiny which are responsible for the development of Gautam Buddha Nagar, India, 2023, pp. 1283-1288,
the face synthesis technology by emphasising on the doi: 10.1109/IC3I59117.2023.10397956
importance of potential biases and moral implications [11]. Sarmah, J., Saini, M.L., Kumar, A., Chasta, V. (2024).
investigation. Altogether, the approach presented in the paper Performance Analysis of Deep CNN, YOLO, and
pave a way for continuing progress in the face synthesis LeNet for Handwritten Digit Classification. In: Sharma,
application across various fields such as healthcare, forensics, H., Chakravorty, A., Hussain, S., Kumari, R. (eds)
entertainment industry etc. Artificial Intelligence: Theory and Applications. AITA
2023. Lecture Notes in Networks and Systems, vol 844.
REFERENCES Springer, Singapore. https://doi.org/10.1007/978-981-
99-8479-4_16
[1]. Pranoto, H., Heryadi, Y., Warnars, H. L. H. S., & [12]. M. L. Saini, B. Tripathi, J. Kaushal and A. Garg, "A
Budiharto, W. (2022). Recent generative adversarial Hybrid Model for Diagnosis Cardiovascular Disease
approach in face aging and dataset review. IEEE Using Clinical Features, ECG and MRI," 2023
Access, 10, 28693-28716. International Conference on Ambient Intelligence,
[2]. Antipov, G., Baccouche, M., & Dugelay, J. L. (2017, Knowledge Informatics and Industrial Electronics
September). Face aging with conditional generative (AIKIIE), Ballari, India, 2023, pp. 1-6, doi:
adversarial networks. In 2017 IEEE international 10.1109/AIKIIE60097.2023.10390299.
conference on image processing (ICIP) (pp. 2089-
2093). IEEE.

IJISRT24AUG265 www.ijisrt.com 699

Volume 9, Issue 8, August – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24AUG265

[13]. M. Lal Saini, B. Tripathi and M. S. Mirza, "Evaluating

the Performance of Deep Learning Models in
Handwritten Digit Recognition," 2023 3rd International
Conference on Technological Advancements in
Computational Sciences (ICTACS), Tashkent,
Uzbekistan, 2023, pp. 116-121, doi:
10.1109/ICTACS59847.2023.10390027.
[14]. Chopra and M. Lal Saini, "Comparison Study of
Different Neural Network Models for Assessing
Employability Skills of IT Graduates," 2023
International Conference on Sustainable
Communication Networks and Application (ICSCNA),
Theni, India, 2023, pp. 189-194, doi:
10.1109/ICSCNA58489.2023.10368605.
[15]. E. G. Kumar, M. Lal Saini, S. A. Khadar Ali and B. B.
Teja, "A Clinical Support System for Prediction of
Heart Disease using Ensemble Learning
Techniques," 2023 International Conference on
Sustainable Communication Networks and Application
(ICSCNA), Theni, India, 2023, pp. 926-931, doi:
10.1109/ICSCNA58489.2023.10370569.
[16]. S. P. Kumar Mygapula, M. Lal Saini and C. S. Raj
Dheeraj, "Performance Evaluation of Machine Learning
Algorithms for Prediction of Cardiac Failure," 2023
International Conference on Sustainable
Communication Networks and Application (ICSCNA),
Theni, India, 2023, pp. 1599-1604, doi:
10.1109/ICSCNA58489.2023.10368606.
[17]. S. Chalechema, M. L. Saini, I. Perla and A. V.
Shivanand, "Customer Segmentation Using K Means
Algorithm and RFM Model," 2023 International
Conference on Computing, Communication, and
Intelligent Systems (ICCCIS), Greater Noida, India,
2023, pp. 393-398, doi:
10.1109/ICCCIS60361.2023.10425556.
[18]. S. Mittal, R. Agarwal, M. L. Saini and A. Kumar, "A
Logistic Regression Approach for Detecting Phishing
Websites," 2023 International Conference on Advances
in Computation, Communication and Information
Technology (ICAICCIT), Faridabad, India, 2023, pp.
76-81, doi: 10.1109/ICAICCIT60255.2023.10466221.
[19]. V. Prabhas, M. Lal Saini, C. Mohith, R. Kumar and B.
Tripathi, "Segmentation of E-Commerce Data Using K-
Means Clustering Algorithm," 2023 Global Conference
on Information Technologies and Communications
(GCITC), Bangalore, India, 2023, pp. 1-6, doi:
10.1109/GCITC60406.2023.10426132.
[20]. M. L. Saini, A. Patnaik, Mahadev, D. C. Sati and R.
Kumar, "Deepfake Detection System Using Deep
Neural Networks," 2024 2nd International Conference
on Computer, Communication and Control (IC4),
Indore, India, 2024, pp. 1-5, doi:
10.1109/IC457434.2024.10486659.

IJISRT24AUG265 www.ijisrt.com 700