Read Anomaly Detection Using a Variational Autoencoder Neural Network With a Novel Objective Function and Gaussian Mixture Model Selection Technique by Naval Postgraduate School Online

Anomaly Detection Using a Variational Autoencoder Neural Network With a Novel Objective Function and Gaussian Mixture Model Selection Technique

Naval Postgraduate School

0/5 ( ratings)

Anomalies in data often convey critical information that can be leveraged in a variety of applications. For the military engaged in combat, this can amount to identifying threats early and preserving a lethal edge over an adversary. In other more benign cases it can corrupt data integrity and lead to ineffective application of other data analysis techniques. To tackle the problem of anomaly detection, there are several common methods provided in statistics and machine learning literature, including variational autoencoders . Using a VAE, we develop a novel objective function to improve its performance detecting anomalies. Additionally, we introduce a modeling pipeline that works in the fully unsupervised context, where one does not know the true proportion of anomalies present in the data. To construct this pipeline, we fit reconstruction errors using a Gaussian mixture model and select the model whose characteristics best match our performance metrics. Using our approach, we observe an increase in anomalies detected against a standard objective function, and we measure an average improvement of 0.4021 in F1 scores. We show our findings using four labeled benchmark data sets and apply our conclusions on an open-source, unlabeled data set taken from USASpending.gov.

Pages: 81

Format: Kindle Edition

Release: August 19, 2019

Rate this book!

Write a review?

Subscribe to Read | $0.00

Popular Genres

Subscribe to Read | $0.00

Anomaly Detection Using a Variational Autoencoder Neural Network With a Novel Objective Function and Gaussian Mixture Model Selection Technique

Anomaly Detection Using a Variational Autoencoder Neural Network With a Novel Objective Function and Gaussian Mixture Model Selection Technique

Anomaly Detection Using a Variational Autoencoder Neural Network With a Novel Objective Function and Gaussian Mixture Model Selection Technique

More books from Naval Postgraduate School

Preventing School Shootings: A Public Health Approach to Gun Violence

Combatting Terrorism Through Study of the Genetic Psychology of Terrorist Leaders: The Early Development of the Terrorist Mind

The Strategic Defense Initiative and the End of the Cold War - SDI, Star Wars, President Reagan, Fall of the Soviet Union, Gorbachev, KAL 007, MAD Nuclear Policy, Teller, ABM Treaty, Reykjavik Summit

The Illusion of Control: Great Powers Interacting with Tribal Societies and Weak Nation-States (French and Indian War Book 1)

Combatting Terrorism Through Study of the Genetic Psychology of Terrorist Leaders: The Early Development of the Terrorist Mind

Quality of Service and Cybersecurity Communication Protocols Analysis for the Robot Operating System 2

Triggers of Violence in New Religious Movements

Al-Ikhwan Al Muslimeen: The Muslim Brotherhood

Multidisciplinary Perspectives on Military Deception

NATO: Maintaining Relevance in the Twenty-First Century

Venezuelan Bolivarian Missions in Colombia: What are the Real, Underlying Reasons for Their Existence

Military Innovation in the Rise and Fall of Great Powers: Lessons for America

The Silent Revolution Within NATO Logistics: A Study in Afghanistan Fuel and Future Applications (Defense)

The Role of Cultural Understanding and Language Training in Unconventional Warfare

Winning the Strategic Narrative in the Israeli-Palestinian Protracted Conflict

Special Operations Liaison Officer: Looking Back to See the Future

The String of Pearls: Chinese Maritime Presence in the Indian Ocean and its Effect on Indian Naval Doctrine

Deradicalizing and Disengaging the Children of the Islamic State

Spaced-Based Solar Power System Architecture

German Views of Irregular Warfare - Hessian Officer Johann Ewald, Carl von Clausewitz, Paul von Lettow-Vorbeck, Otto Skorzeny in World War II, Commando Raid, Friedrich von der Heydte, Otto Heilbrunn