Citation :

Mohit Jain, Adit Shah, 2024. "Anomaly Detection Using Convolutional Neural Networks (CNN)" ESP International Journal of Advancements in Computational Technology (ESP-IJACT)  Volume 2, Issue 3: 12-22.

Abstract :

This paper targets detection of anomalies and it is a significant process in different disciplines including fraud detection, network security, industrial inspection, and health care. The previous methods of detecting anomalies depend on statistics or using methods of classical machine learning, which are inefficient in recognizing complicated patterns of different high-dimensional spaces. Deep learning enabled feature extraction and pattern recognition in CNN, benefiting from the robust computational power of the networks that make it ideal for use in anomaly detection. This paper aims at reviewing the current state of the art and knowledge concerning the use of CNNs for anomaly detection and the techniques, methodologies, and results that the current literature offers on the subject. It also explores the various types of anomalies, the structure and design of specific single-shot CNNs appropriate for the detection of anomalies, and the ways of evaluating performance. Moreover, functional difficulties and prospects in this field are far from obvious. Thus, this paper will engage in a literature review and systematic approach to offer an understanding of CNNs and its performance in recognizing anomalies in diverse applications.

References :

[1] Convolutional Neural Networks, Explained, Medium. https://towardsdatascience.com/convolutional-neural-networks-explained-9cc5188c4939

[2] Deep Learning for Anomaly Detection, Cloudera Fast Forward. https://ff12.fastforwardlabs.com/

[3] Ajitesh Kumar, Different Types of CNN Architectures Explained: Examples, Vitalflux. https://vitalflux.com/different-types-of-cnn-architectures-explained-examples/

[4] Convolutional (CNN/CNN)-based Encoder-Decoder Neural Network, Gabormelli. https://www.gabormelli.com/RKB/Convolutional_(CNN/CNN)-based_Encoder-Decoder_Neural_Network

[5] Generative Adversarial Network (GAN), geeksforgeeks. https://www.geeksforgeeks.org/generative-adversarial-network-gan/

Keywords :

Anomaly Detection, Convolutional Neural Networks (CNN), Deep Learning, Feature Extraction, Industrial Inspection.