eprintid: 20260
rev_number: 2
eprint_status: archive
userid: 1
dir: disk0/00/02/02/60
datestamp: 2024-06-04 14:20:00
lastmod: 2024-06-04 14:20:00
status_changed: 2024-06-04 14:17:05
type: book
metadata_visibility: show
creators_name: Naveed, F.
creators_name: Imran, S.A.
creators_name: Janisar, A.A.
creators_name: Yusuf, A.
creators_name: Khan, S.
creators_name: Khan, I.U.
title: AI Classification Algorithms for Human Activities Recognition System With a Cyber Security Perspective
ispublished: pub
note: cited By 0
abstract: Today�s world is heavily reliant on electronic technology, which attracts Cyber attacks and threats to the cyber security systems. These attacks and vulnerabilities need to be overcome using cyber security and advanced machine learning mechanisms together. This study concentrates on human activity recognition (HAR) problems. A novel method is proposed for recognizing human activity with the assistance of different calculations in order to pick the precise one. The research utilizes HARs rather than human activity recognition from consistency. HAR, the computerized revelation of advancing activities from visual data, is a prominent issue. Semantic assessment of activity chronicles engages the improvement of various vision-dependent on structures, including shrewd observation systems, smart robots, activity-based human-PC interfaces, and noticing systems for individuals. For instance, a way to deal with normally perceived and know the entire class of significant exercises. A number of studies were performed on various joint efforts in a comparable scene, to figure out briefly and to spatially shape complex critical-level activities of a reformist nature. There are different methods to perceive HAR from different sources; this study assures the accuracy of different calculations and then conceives the outcomes utilizing a few lattices like accuracy, precision, recall, and F-measure. The dataset is downloaded from a UCI, and the data is based on a sensor-based gyroscope/accelerometer. Different techniques are implemented to develop a recognition system to build a HAR system; however, this method utilized a dataset to generate possible/precise outcomes. Experiments were carried out on the preprocessing features. SVM, ANN, and K-NN are examples of classifiers and selection methods used to get the desired results. Among all the classifiers, SVM produced the highest result in the dataset, 97.895. In the dataset, K-NN and decision tree produced the same result, 93.64. Dataset decision tree yielded the low result of 93, while SVM produced a result of 97.895. Despite having a vast AI environment, cyber attacks had the largest attack surface. A cyber attack would be catastrophic in future for businesses continuity if the AI perspective is missing in it. © 2024 selection and editorial matter, Inam Ullah Khan, Mariya Ouaissa, Mariyam Ouaissa, Zakaria Abou El Houda and Muhammad Fazal Ijaz; individual chapters, the contributors.
date: 2024
publisher: CRC Press
official_url: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85179238236&doi=10.1201%2f9781003404361-16&partnerID=40&md5=9eb7a07b87a0d32d0f35ea8e4e8f0a04
id_number: 10.1201/9781003404361-16
full_text_status: none
publication: Cyber Security for Next-Generation Computing Technologies
pagerange: 294-313
refereed: TRUE
isbn: 9781003826408; 9781032518992
citation:   Naveed, F. and Imran, S.A. and Janisar, A.A. and Yusuf, A. and Khan, S. and Khan, I.U.  (2024) AI Classification Algorithms for Human Activities Recognition System With a Cyber Security Perspective.     CRC Press, pp. 294-313.  ISBN 9781003826408; 9781032518992