{"678314":{"#nid":"678314","#data":{"type":"event","title":"PhD Proposal by Matthew Lau","body":[{"value":"\u003Cp\u003E\u003Cstrong\u003ETitle\u003C\/strong\u003E: Characterizing Anomalies for Reliable Machine Learning\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003EDate\u003C\/strong\u003E: Monday, November 18, 2024\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003ETime\u003C\/strong\u003E: 9:30 AM \u2013 10:45 AM ET\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003ELocation\u0026nbsp;\u003C\/strong\u003E[Hybrid]:\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E- Coda C0908 Home Park\u003C\/p\u003E\u003Cp\u003E- Zoom Link: \u003Ca href=\u0022https:\/\/www.google.com\/url?q=https:\/\/gatech.zoom.us\/j\/92792928638?pwd%3DkkeTGxpYxd7bu3zZ2cx0XZ5PS5o918.1\u0026amp;sa=D\u0026amp;source=calendar\u0026amp;usd=2\u0026amp;usg=AOvVaw3jbtyYDExBcM-NUkVc54I7\u0022\u003E\u003Cem\u003Ehttps:\/\/gatech.zoom.us\/j\/92792928638?pwd=kkeTGxpYxd7bu3zZ2cx0XZ5PS5o918.1\u003C\/em\u003E\u003C\/a\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003EMatthew Lau\u003C\/strong\u003E\u003C\/p\u003E\u003Cp\u003EPh.D. CS Student\u003C\/p\u003E\u003Cp\u003ESchool of Cybersecurity and Privacy\u003C\/p\u003E\u003Cp\u003ECollege of Computing\u003C\/p\u003E\u003Cp\u003EGeorgia Institute of Technology\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003ECommittee:\u003C\/strong\u003E\u003C\/p\u003E\u003Cp\u003EDr. Wenke Lee - (Advisor) School of Cybersecurity and Privacy, Georgia Institute of Technology\u003C\/p\u003E\u003Cp\u003EDr. Athanasios P. (Sakis) Meliopoulos - School of Electrical and Computer Engineering, Georgia Institute of Technology\u003Cbr\u003EDr. Saman Zonouz - School of Cybersecurity and Privacy\u0026nbsp;\u0026amp;\u0026nbsp;School of Electrical and Computer Engineering,\u0026nbsp;Georgia Institute of Technology\u003C\/p\u003E\u003Cp\u003EDr.\u0026nbsp;(Polo)\u0026nbsp;Chau Duen Horng - School of Computational Science \u0026amp; Engineering, Georgia Institute of Technology\u003C\/p\u003E\u003Cp\u003EDr. Huo Xiaoming - School of Industrial and Systems\u0026nbsp;Engineering, Georgia Institute of Technology\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003EAbstract:\u003C\/strong\u003E\u003C\/p\u003E\u003Cp\u003EMachine learning (ML) has had much success across a variety of domains and tasks over the past few decades. However, ML models often assume that test data statistically mirrors the training data, an assumption that fails in the presence of anomalies (i.e., test data that do not mirror training). Yet, scenarios that produce anomalies are precisely the situations that can be safety- and security-critical, such as cyber-attacks. To ensure that ML models are reliable (i.e., they are accurate \u003Cem\u003Eeven\u0026nbsp;\u003C\/em\u003Ewhen fed anomalies), we propose a framework to characterize anomalies and incorporate this characterization into the ML pipeline. We discuss how to use this framework for unknown and foreseeable types of anomalies, both of which we have no data for.\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003EFor unknown anomalies, we characterize them as living in large open spaces and ensure that models are conservative in these open spaces. The presence of unknown anomalies is a key trait of anomaly detection. In these large open spaces, we bias neural networks to be conservative, classifying open spaces as anomalous. We show how to bias neural networks statistically and geometrically for unsupervised and supervised anomaly detection respectively. With this bias, we improve the reliability of neural networks on unknown anomalies.\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003EFor foreseeable anomalies, we aim to analyze and account for their pattern (known as \u003Cem\u003Esignature\u003C\/em\u003E) by feature engineering. Attacks on cyber-physical systems (CPSes) are anomalies that we can foresee due to attacks being constrained by the cyber or physical component. Here, we characterize each attack signature and ensure that the ML model accounts for it. We show that our approach is principled by evaluating with two case studies on (1) cyber-attacks against explainable anomaly detection on power grids and (2) physical adversarial attacks against video-based object detection. In the first case study, we design graph change statistics to localize attacked sensors with phase- and amplitude-based signatures. For the second case study, we project images onto the data manifold with background subtraction for model fine-tuning, promoting robustness against off- and on-manifold adversarial signatures. In these two cases, characterizing attack signatures with feature engineering ensures that ML models are accurate even during attacks.\u0026nbsp;\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003EIn summary, this thesis proposes a framework to characterize anomalies in ML. For unknown anomalies, we encourage ML models to be conservative in large open spaces. When more information is present, we can use feature engineering to account for signatures from foreseeable anomalies. Accounting for potential anomalies in both cases, we increase the reliability of ML.\u003C\/p\u003E","summary":"","format":"limited_html"}],"field_subtitle":"","field_summary":[{"value":"\u003Cp\u003ECharacterizing Anomalies for Reliable Machine Learning\u003C\/p\u003E","format":"limited_html"}],"field_summary_sentence":[{"value":"Characterizing Anomalies for Reliable Machine Learning"}],"uid":"27707","created_gmt":"2024-11-11 15:00:21","changed_gmt":"2024-11-11 15:00:21","author":"Tatianna Richardson","boilerplate_text":"","field_publication":"","field_article_url":"","field_event_time":{"event_time_start":"2024-11-18T09:30:40-05:00","event_time_end":"2024-11-18T11:00:00-05:00","event_time_end_last":"2024-11-18T11:00:00-05:00","gmt_time_start":"2024-11-18 14:30:40","gmt_time_end":"2024-11-18 16:00:00","gmt_time_end_last":"2024-11-18 16:00:00","rrule":null,"timezone":"America\/New_York"},"location":"- Coda C0908 Home Park","extras":[],"groups":[{"id":"221981","name":"Graduate Studies"}],"categories":[],"keywords":[{"id":"102851","name":"Phd proposal"}],"core_research_areas":[],"news_room_topics":[],"event_categories":[{"id":"1788","name":"Other\/Miscellaneous"}],"invited_audience":[{"id":"78771","name":"Public"}],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[],"email":[],"slides":[],"orientation":[],"userdata":""}}}