Anomaly Detection

By S. Mukherjee. ACM SIGSPATIAL International Workshop on Advances in Geographic Information Systems, 2019

653 citations recorded by the source corpus.

Due to the significance it holds in the concept of fraud, security in computers and business, anomaly detection serves very much purpose. Using techniques in unsupervised machine learning, the two algorithms applied in this study are Isolation Forest and Autoencoder in credit card fraud detection in financial datasets. This work focuses on data preparation and selection, generation and extraction of the features, as well as model assessment through use of metrics such as ROC-AUC, measure of precision, and measure of recall. It also discovered that Autoencoder attends to complex patterns of anomalies while Isolation Forest cuts down the false positives. Some of these problems include class imbalance and computational issues are highlighted. Some of these strategies include hybridization for imbalance handling and real time implementation that is very helpful in the development of automated and large scale anomaly detection in financial related work.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/ff186d771a6ad1c5900fae3b0ac342aafbd053e3

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

GANomaly: Semi-Supervised Anomaly Detection via Adversarial Training (2018, background): “Besides, a considerable amount of work taxonomized the approaches within the literature [9,19,28,29,33].”
A Multimodal Anomaly Detector for Robot-Assisted Feeding Using an LSTM-Based Variational Autoencoder (2017, background): “Anomaly detection is known as novelty, outlier, or event detections [6].”
Fraud detection system: A survey (2016, methodology): “Pro ﬁ le approaches can be done based on owner approach or operation approach as classi ﬁ ed by Chandola et al. (2009).”