EXPOSING DEEPFAKES IN SOCIAL MEDIA PLATFORMS

USING GENERATIVE ADVERSARIAL NETWORKS

Mrs.Chithra.B MCA
Nandhini.S ,Priyadharshini.D
Dept of CSE
PERI Institute of technology,Chennai.
Abstract-This project delves into the critical realm of combating deep fakes in
online networking platforms by employing advanced deep learning techniques.
Leveraging Generative Adversarial Networks (GANs) to simulate deep fake
generation and utilizing the feature extraction capabilities of Inception
ResNetV2, the research aims to develop a robust deep fake detection model. The
proposed model undergoes comprehensive training, evaluation, and fine-tuning,
with a focus on countering adversarial techniques employed in sophisticated deep
fake generation. The study contributes a reliable and effective means of
discerning between genuine and synthetic content, offering heightened security
for online networking platforms. Furthermore, insights gained into adversarial
strategies and practical deployment recommendations provide a comprehensive
approach to addressing the rising threat of deep fakes in the digital landscape.The
proposed modelundergos comprehensive training,evaluation,and fine-
tunining,with focus on countering adversarial techniques employed in
sophisticated fpdeep fake generation.

1 INTRODUCTION
The expanding complexity of cell phone cameras and the accessibility of
good web association all around the world has expanded the always developing
reach of online media also, media sharing entries have made the creation and
transmission of computerized recordings more simple than any time in recent
memory. The developing computational force has made deep learning so
incredible that would have been thought unthinkable just a modest bunch of years
prior. Like any extraordinary innovation, this has made new difficulties.
Purported "DeepFake" created by deep generative adversarial models that can
control video and brief snippets. Spreading of the DF over the online media stages
have gotten extremely normal prompting spamming and peculating incorrect data
over the stage. These sorts of the DF will be awful, and lead to threatening,
deluding of average citizens. To conquer this sort of situation, DF detection is
very essential. So, we describe a brand new deep learningbased approach that
could successfully distinguish AI-generated fake videos (DF Videos) from actual
videos. It’s incredibly essential to broaden technology that could spot fakes, so
that the DF may be recognized and averted from spreading over the internet.

2 EXISTING SYSTEM
The detection of digital face manipulation in video has attracted
extensive attention due to the increased risk to public trust. To counteract the
malicious usage of such techniques, deep learning-based deepfake detection
methods have been developed and have shown impressive results. However, the
performance of these detectors is often evaluated using benchmarks that hardly
reflect real-world situations. For example, the impact of various video processing
operations on detection accuracy has not been systematically assessed. To
address this gap, this project first analyses numerous real-world influencing
factors and typical video processing operations. Then, a more systematic
assessment methodology is proposed, which allows for a quantitative evaluation
of a detector’s robustness under the influence of different processing operations.
For this entire process they used the SVM model.

3 PROPOSED SYSTEM
Increasing computing power has made deep learning algorithms so
powerful that creating a fake video generated by artificial intelligence, popularly
called as deep fakes, is very simple. Scenarios where this realistic face has been
replaced by deep fakes are used to create political unrest, fake terrorist acts,
revenge porn, blackmailing nations can easily be imagined. In this work, a new
method is based on deep learning that can effectively distinguish fake videos
generated by artificial intelligence from real videos. This method is able to
automatically detect replacement and reenactment of deep forgery. Using
artificial intelligence (AI) to fight against artificial intelligence (AI). This system
uses ResNext Convolution Neural Network to extract frame-level features and
these features and further uses GAN and ResNetV2 to classify whether the video
is subject to some kind of manipulation or not, i.e.. whether the video is deepfake
or real video. System can also achieve competitive results using a very simple
and robust approach.

4 REQUIREMENTS
There are two kinds of Requirements.without these requirements and tools
project can be done.
They are
o Hardware Requirements
o Software Requirements
4.1 Hardware requirements:
✓ HDD: >90GB
✓ PROCESSOR: >Pentium IV 2.4GHz
✓ SYSTEM TYPE: 32bit / 64 bit
✓ RAM: >2GB
✓ OS: WINDOWS 7/8/8.1/10

4.2 Software requirements:

✓ TOOL : MATLAB 2019a
✓ TOOL BOX : Image Processing Tool Box

5 SYSTEM ARCHITECTURE
5.1 ARCHITECTURE EXPLANATION

A convolutional neural network (CNN or ConvNet is a network architecture

for deep learning that learns directly from the data below in Fig. CNNs are useful
for finding image patternsto recognize objects, classes, and categories.
Convolutional Neural Networks comprise node layers, convolution, pooling,
hidden, and output layers. Our image passes through all these layers, and every
step is important. As for the convolution layer, this layer is the main block of
CNN. This is the layer where most of the computations occur. This layer has three
inputs. The first is input data, the second is a filter, and the third is a feature map.
The convolution process consists of the input image used with filters, and filters
are mainly 3x3 matrices that are iterated over the image. In the end, an activation
function calculates the final result. The output is stored in an output matrix. The
pooling layer is used for downsampling, reducing the number of parameters in
our input image. The fully connected layer is the layer that performs classification
on the image and decides whether an image is real or fake.

6 CONCLUSION
We provided a neural network-primarily based totally method to
classify the video as deep fake or actual, at the side of the self-assurance of the
proposed model. The proposed approach is stimulated with the aid of using the
manner the deep fakes are created with the aid of using the GANs with the assist
of Autoencoders. Our approach does the frame stage detection the use of
ResNetV2 and video class the use of RNN at the side of GAN. The proposed
approach is successful in detecting the video as a deep fake or actual primarily
based totally on the listed parameters in the paper. We consider that it'll offer a
very excessive accuracy on actualtime data.

7 FUTURE ENHANCEMENT
The detection showed its best performance with the CNN, and more
accurate results are aimed to be established in the future for a more authentic,
precise website. It is also intended to work with different datasets and generalize
the dataset more with several augmentation techniques so that the system can
detect any data inserted by the user.

REFERENCES
1) T. Mahara, V. L. H. Josephine, R. Srinivasan, P. Prakash, A. D. Algarni
and O. P. Verma, "Deep vs. Shallow: A Comparative Study of Machine
Learning and Deep Learning Approaches for Fake Health News
Detection," in IEEE Access, vol. 11, pp. 79330-79340, 2023, doi:
10.1109/ACCESS.2023.3298441.
2) Z. M. Almutairi and H. Elgibreen, "Detecting Fake Audio of Arabic
Speakers Using Self-Supervised Deep Learning," in IEEE Access, vol. 11,
pp. 72134-72147, 2023, doi: 10.1109/ACCESS.2023.3286864.
3) A. A. Obaid, H. Khotanlou, M. Mansoorizadeh and D. Zabihzadeh,
"Robust Semi-Supervised Fake News Recognition by Effective
Augmentations and Ensemble of Diverse Deep Learners," in IEEE Access,
vol. 11, pp. 54526-54543, 2023, doi: 10.1109/ACCESS.2023.3278323.
4) M. Tajrian, A. Rahman, M. A. Kabir and M. R. Islam, "A Review of
Methodologies for Fake News Analysis," in IEEE Access, vol. 11, pp.
73879-73893, 2023, doi: 10.1109/ACCESS.2023.3294989.
5) K. Li, X. Lu, M. Akagi and M. Unoki, "Contributions of Jitter and Shimmer
in the Voice for Fake Audio Detection," in IEEE Access, vol. 11, pp.
84689-84698, 2023, doi: 10.1109/ACCESS.2023.3301616.
6) B. L. V. S. Aditya and S. N. Mohanty, "Heterogenous Social Media
Analysis for Efficient Deep Learning Fake-Profile Identification," in IEEE
Access, vol. 11, pp. 99339-99351, 2023, doi:
10.1109/ACCESS.2023.3313169.
7) N. M. Alnaim, Z. M. Almutairi, M. S. Alsuwat, H. H. Alalawi, A.
Alshobaili and F. S. Alenezi, "DFFMD: A Deepfake Face Mask Dataset
for Infectious Disease Era With Deepfake Detection Algorithms," in IEEE
Access, vol. 11, pp. 16711-16722, 2023, doi:
10.1109/ACCESS.2023.3246661.
8) R. Catelli et al., "A New Italian Cultural Heritage Data Set: Detecting Fake
Reviews With BERT and ELECTRA Leveraging the Sentiment," in IEEE
Access, vol. 11, pp. 52214-52225, 2023, doi:
10.1109/ACCESS.2023.3277490.
9) M. O. Alassafi et al., "A Novel Deep Learning Architecture With Image
Diffusion for Robust Face Presentation Attack Detection," in IEEE Access,
vol. 11, pp. 59204-59216, 2023, doi: 10.1109/ACCESS.2023.3285826.
10) M. Rajaee and K. Mazlumi, "Multi-Agent Distributed Deep
Learning Algorithm to Detect Cyber-Attacks in Distance Relays," in IEEE
 Access, vol. 11, pp. 10842-10849, 2023, doi:
10.1109/ACCESS.2023.3239684.
11) D. Benalcazar, J. E. Tapia, S. Gonzalez and C. Busch, "Synthetic ID
Card Image Generation for Improving Presentation Attack Detection," in
IEEE Transactions on Information Forensics and Security, vol. 18, pp.
1814-1824, 2023, doi: 10.1109/TIFS.2023.3255585.
12) A. H. Khalil, A. Z. Ghalwash, H. A. -G. Elsayed, G. I. Salama and
H. A. Ghalwash, "Enhancing Digital Image Forgery Detection Using
Transfer Learning," in IEEE Access, vol. 11, pp. 91583-91594, 2023, doi:
10.1109/ACCESS.2023.3307357.
13) I. -Y. Kwak et al., "Voice Spoofing Detection Through Residual
Network, Max Feature Map, and Depthwise Separable Convolution," in
IEEE Access, vol. 11, pp. 49140-49152, 2023, doi:
10.1109/ACCESS.2023.3275790.
14) H. Alamro, K. Mahmood, S. S. Aljameel, A. Yafoz, R. Alsini and
A. Mohamed, "Modified Red Fox Optimizer With Deep Learning Enabled
False Data Injection Attack Detection," in IEEE Access, vol. 11, pp.
79256-79264, 2023, doi: 10.1109/ACCESS.2023.3298056.
15) R. M. Albalawi, A. T. Jamal, A. O. Khadidos and A. M. Alhothali,
"Multimodal Arabic Rumors Detection," in IEEE Access, vol. 11, pp.
9716-9730, 2023, doi: 10.1109/ACCESS.2023.3240373.
16) R. A. Zayed, L. F. Ibrahim, H. A. Hefny, H. A. Salman and A.
AlMohimeed, "Experimental and Theoretical Study for the Popular
Shilling Attacks Detection Methods in Collaborative Recommender
System," in IEEE Access, vol. 11, pp. 79358-79369, 2023, doi:
10.1109/ACCESS.2023.3289404.
17) A. A. Shafee, M. M. E. A. Mahmoud, G. Srivastava, M. M. Fouda,
M. Alsabaan and M. I. Ibrahem, "Detection of Distributed Denial of
Charge (DDoC) Attacks Using Deep Neural Networks With Vector
 Embedding," in IEEE Access, vol. 11, pp. 75381-75397, 2023, doi:
10.1109/ACCESS.2023.3296562.
18) Y. Patel et al., "An Improved Dense CNN Architecture for Deepfake
Image Detection," in IEEE Access, vol. 11, pp. 22081-22095, 2023, doi:
10.1109/ACCESS.2023.3251417.
19) Z. Liu, J. Hu, Y. Liu, K. Roy, X. Yuan and J. Xu, "Anomaly-Based
Intrusion on IoT Networks Using AIGAN-a Generative Adversarial
Network," in IEEE Access, vol. 11, pp. 91116-91132, 2023, doi:
10.1109/ACCESS.2023.3307463.
20) J. M. Pérez et al., "Assessing the Impact of Contextual Information
in Hate Speech Detection," in IEEE Access, vol. 11, pp. 30575-30590,
2023, doi: 10.1109/ACCESS.2023.3258973.
21) M. J. Abdulaal et al., "Privacy-Preserving Detection of Power Theft
in Smart Grid Change and Transmit (CAT) Advanced Metering
Infrastructure," in IEEE Access, vol. 11, pp. 68569-68587, 2023, doi:
10.1109/ACCESS.2023.3291217.
22) S. Asiri, Y. Xiao, S. Alzahrani, S. Li and T. Li, "A Survey of
Intelligent Detection Designs of HTML URL Phishing Attacks," in IEEE
Access, vol. 11, pp. 6421-6443, 2023, doi:
10.1109/ACCESS.2023.3237798.
23) W. M. B. Ateeq and H. S. Al-Khalifa, "Intelligent Framework for
Detecting Predatory Publishing Venues," in IEEE Access, vol. 11, pp.
20582-20618, 2023, doi: 10.1109/ACCESS.2023.3250256.
24) S. Y. Diaba, M. Shafie-Khah and M. Elmusrati, "Cyber Security in
Power Systems Using Meta-Heuristic and Deep Learning Algorithms," in
IEEE Access, vol. 11, pp. 18660-18672, 2023, doi:
10.1109/ACCESS.2023.3247193.
25) R. Sepúlveda-Torres, A. Bonet-Jover and E. Saquete, "Detecting
Misleading Headlines Through the Automatic Recognition of
Contradiction in Spanish," in IEEE Access, vol. 11, pp. 72007-72026,
2023, doi: 10.1109/ACCESS.2023.3295781.