A Performance Enhancement of Deepfake Video Detection through the use of a Hybrid CNN Deep Learning Model

Ikram, Sumaiya Thaseen; Priya, V.; Chambial, Shourya; Sood, Dhruv; V, Arulkumar

doi:10.32985/ijeces.14.2.6

Cited by 15 publications

(3 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In 2018 [28] have suggested a hybrid IDS via Bayesian, Associative, and Decision tree approaches to protect the data from intrusion. The experimentation involved sensors from IDS on each Cloud host computer.…”

Section: Literature Review Related Workmentioning

confidence: 99%

“…The ensemble-based ML method [26] enhanced accuracy and the minimum error still, computation complexity due to multiple approaches limits the performance. The Bayesian, Associative, and Decision tree approaches [28] obtained considerable performance and robustness while for real-time experimentation, it expressed moderate performance. The Heroku with AES model assured better data protection yet, there is no surety about the accuracy of data security.…”

Section: Reviewmentioning

confidence: 99%

“…ShiftRows: The state's four rows of bytes have been cylindrically moved to the (𝑏𝑏 − 1) position, in which 𝑏𝑏refers to the number of rows between1 𝐹𝐹𝑝𝑝 4. MixColumn: A Galois Field (GF) multiplication is used to multiply 4 × 4 matrix various columns and a GF (28) multiplication is employed to attain column-by-column multiplication. AddRoundkey: An X-OR operation performs with add round key and the present state to the state.…”

Section: Security Architecture Traditional Aes Algorithmmentioning

confidence: 99%

See 2 more Smart Citations

LM-GA: A Novel IDS with AES and Machine Learning Architecture for Enhanced Cloud Storage Security

Thilagam

Aruna

2023

JMC

View full text Add to dashboard Cite

Cloud Computing (CC) is a relatively new technology that allows for widespread access and storage on the internet. Despite its low cost and numerous benefits, cloud technology still confronts several obstacles, including data loss, quality concerns, and data security like recurring hacking. The security of data stored in the cloud has become a major worry for both Cloud Service Providers (CSPs) and users. As a result, a powerful Intrusion Detection System (IDS) must be set up to detect and prevent possible cloud threats at an early stage. Intending to develop a novel IDS system, this paper introduces a new optimization concept named Lion Mutated-Genetic Algorithm (LM-GA) with the hybridization of Machine Learning (ML) algorithms such as Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM). Initially, the input text data is preprocessed and balanced to avoid redundancy and vague data. The preprocessed data is then subjected to the hybrid Deep Learning (DL) models namely the CNN-LSTM model to get the IDS output. Now, the intruded are discarded and non-intruded data are secured using Advanced Encryption Standard (AES) encryption model. Besides, the optimal key selection is done by the proposed LM-GA model and the cipher text is further secured via the steganography approach. NSL-KDD and UNSW-NB15 are the datasets used to verify the performance of the proposed LM-GA-based IDS in terms of average intrusion detection rate, accuracy, precision, recall, and F-Score.

show abstract

Section: Literature Review Related Workmentioning

confidence: 99%

Section: Reviewmentioning

confidence: 99%

Section: Security Architecture Traditional Aes Algorithmmentioning

confidence: 99%

See 1 more Smart Citation