Special Issue(24-20): Novel Machine Learning Methods for Anomaly Detection in Sensor Networks Posted on 2024-08-27

Novel Machine Learning Methods for Anomaly Detection in Sensor Networks

 

   Machine learning has become a potential tool for anomaly identification. Without the requirement for manual rule development, machine learning algorithms are able to learn from sensor data and identify patterns that indicate both normal and abnormal behaviour. Along with more conventional methods like Isolation Forests and One Class SVMs, promising machine learning techniques for anomaly identification include deep learning techniques like autoencoders and convolutional neural networks. These techniques are excellent for identifying odd patterns in data from a variety of fields. This approach is a tool for anomaly detection in regression modelling or density based classifiers. One statistical method for determining the link between labelled and variable data is regression modelling. HMM and clustering are employed in an unsupervised learning system to identify abnormalities in network traffic. The advantages and disadvantages of each unsupervised approach are discussed in this paper along with the accuracy level. Categorical and continuous data are the primary applications of reinforcement learning. Anomalies are essentially singular, unexpected data items that occur inside a dataset. The words are frequently used synonymously with outliers. Novelties, on the other hand, are anomalies in data that are unique to each new instance. 

 

   The original dataset does not contain them. It is a well known instance based non parametric method for identifying abnormalities. Finding irregularities in the data based on their distance from nearest neighbours is the main objective of this approach. An alternative approach would be to use a rigorous mathematical model to describe the definition of regular system usage and then mark any violation from this model as an attack. The term strict anomaly detection describes this. Additional ways for identifying abnormalities include artificial immune systems, grammar based algorithms, and data mining techniques. Anomaly detection is used in the manufacturing and industrial sector for equipment predictive maintenance. Machine sensors gather information on a range of characteristics, including vibration, sound, and temperature. Using machine learning algorithms, this data is analysed to find trends that might point to equipment breakdowns. 

 

    Novelty detection makes it possible to spot data patterns that noticeably deviate from typical behaviour. It makes it possible to classify and identify items accurately that were unknown to the model throughout its learning phase. Unsupervised anomaly detection is most commonly implemented using artificial neural networks. Because of their layered, non linear learning, ANNs can be trained on vast, unlabeled datasets and are reliable in identifying complex patterns that help classify anomalies of a wide range. Unsupervised anomaly detection algorithms, which can recognize unusual or rare events automatically, may be used in these situations by data scientists. A cloud cost estimator, for instance, can search for odd increases in processing costs or data egress charges that might be the result of a badly constructed algorithm. Reducing human labour, enabling proactive countermeasures, and identifying unexpected events or irregularities in data are some advantages of anomaly based detection. The capacity to identify unknown attacks and adjust to novel attack patterns are two advantages of anomaly based detection. Articles are invited that explore Novel Machine Learning Methods for Anomaly Detection in Sensor Networks. Case studies and practitioner perspectives are also welcome.

 

The topics relevant to this special issue include but are not limited to:

 

  1. Machine learning algorithms for anomaly detection in wireless sensor networks.
  2. Using multi parameterized edit distance and machine learning, wireless sensor networks can detect anomalies over the short and long term.
  3. An investigation on anomaly detection techniques in wireless sensor networks using machine learning.
  4. Machine learning anomaly detection to find IoT system sensor manipulation.
  5. Machine learning for sensor network anomaly assessment for aerospace NDT.
  6. A new adaptive machine learning approach for network anomaly detection. 
  7. A hybrid machine learning approach for identifying anomalies in wireless sensor network unlabeled data.
  8. Enhanced transitory extreme learning system for anomaly detection in wireless sensor networks.
  9. A machine learning strategy based on measurement data for industrial control systems anomaly detection.
  10. Wireless sensor network anomaly detection with ensemble random forest.
  11. Machine learning techniques are being used to detect attacks and anomalies in IoT sensors at IoT locations.
  12. A novel method of employing support vector data description for anomaly detection in wireless sensor networks.



Guest Editor Biographies:

 

Dr. Md. Tarek Habib

Assistant Professor

Department of Computer Science and Engineering

Independent University, Bangladesh

Email: [email protected], [email protected]

Website : http://www.cse.iub.edu.bd/index.php/faculties/73

 

Biography: Md. Tarek Habib achieved his Ph.D. degree at the Department of Computer Science and Engineering at Jahangirnagar University in 2022. He obtained his B.Sc. degree in Computer Science from BRAC University in 2006. Then he got an M.S. degree in Computer Science and Engineering from North South University in 2009. He is an Assistant Professor at the Department of Computer Science and Engineering at Independent University, Bangladesh.

Earlier, he was an Associate Professor at the Department of Computer Science and Engineering

at Daffodil International University. He is much fond of research. He has published many

articles in international journals and conference proceedings. His research interest is in Artificial Intelligence, especially Machine Learning, Artificial Neural Networks, Computer Vision, and Natural Language Processing. Recently, he has started working on IoT too.




Dr. Abdullah Al-Mamun

Assistant Professor

School of Computer and Cyber Science,

Augusta University, Augusta, GA 30912, United States.

Email: [email protected]

Website: https://www.augusta.edu/faculty/directory/view.php?id=AALMAMUN

Biography: Abdullah Al-Mamun is an assistant professor at the Computer and Cyber Sciences

Department at Augusta University. His research interests encompass resource-efficient data

management, distributed systems, and high-performance computing. He collaborates with many industries and academic partners to solve big and significant problems, ranging from HPC infrastructures to large-scale data-intensive computing systems. His research outcomes have been published in premiere venues such as SC, ICDE, IPDPS, NDSS, and ICPP and turned into software prototypes.

 

Dr. Ismail Jabiullah

Professor

Department of Computer Science and Engineering,

Daffodil International University, Birulia, Bangladesh

Email: [email protected]

Website: https://faculty.daffodilvarsity.edu.bd/profile/cse/jabiullah.html

Biography: Md. Ismail Jabiullah is a distinguished academic with over 33 years of experience in teaching, research, and development in Computer Science and Engineering. He serves as a

professor in the Department of Computer Science and Engineering at Daffodil International

University, Birulia, Bangladesh, under the Faculty of Science and Information Technology. His teaching areas include Software Engineering, Cryptography and Network Security, and E-

Commerce. Dr. Jabiullah has authored 27 books, published 78 journal papers, and presented 105 conference papers. His research interests span Network Security, Web Security, Software

Security, Image Processing, Computer Vision, Wireless, Cellular, and Satellite Networks,

Artificial Intelligence, Neural Networks, Software Testing, Machine Learning, and Deep

Learning.

 

Tentative Dates:

Submission Deadline: Dec. 20, 2024

Authors Notification: March. 10, 2025

Revised Papers Deadline: Jun. 10, 2025

Final notification: Sep. 15, 2025