{"629408":{"#nid":"629408","#data":{"type":"event","title":"PhD Proposal by Frederick Hohman","body":[{"value":"\u003Cp\u003E\u003Cstrong\u003ETitle:\u003C\/strong\u003E\u0026nbsp;Interactive Scalable Interfaces for Machine Learning Interpretability\u003Cbr \/\u003E\r\n\u003Cbr \/\u003E\r\nFred Hohman\u003Cbr \/\u003E\r\nSchool of Computational Science \u0026amp; Engineering\u003Cbr \/\u003E\r\nCollege of Computing\u003Cbr \/\u003E\r\nGeorgia Institute of Technology\u003Cbr \/\u003E\r\n\u003Cbr \/\u003E\r\n\u003Cstrong\u003EDate:\u003C\/strong\u003E\u0026nbsp;Monday, December 2, 2019\u003Cbr \/\u003E\r\n\u003Cstrong\u003ETime:\u003C\/strong\u003E\u0026nbsp;4:30pm - 6pm (EST)\u003Cbr \/\u003E\r\n\u003Cstrong\u003ELocation:\u003C\/strong\u003E\u0026nbsp;Coda 114 (first floor conference room)\u003Cbr \/\u003E\r\n\u003Cbr \/\u003E\r\n\u003Cstrong\u003ECommittee\u003C\/strong\u003E\u003Cbr \/\u003E\r\nDuen Horng (Polo) Chau - Advisor, Georgia Tech, Computational Science\u0026nbsp;\u0026amp;\u0026nbsp;Engineering\u003Cbr \/\u003E\r\nAlex Endert - Georgia Tech, Interactive Computing\u003Cbr \/\u003E\r\nChao Zhang - Georgia Tech, Computational Science\u0026nbsp;\u0026amp;\u0026nbsp;Engineering\u003Cbr \/\u003E\r\nNathan Hodas - Pacific Northwest National Lab\u003Cbr \/\u003E\r\nScott Davidoff - NASA Jet Propulsion Lab\u003Cbr \/\u003E\r\nSteven Drucker - Microsoft Research\u003Cbr \/\u003E\r\n\u003Cbr \/\u003E\r\n\u003Cstrong\u003EAbstract\u003C\/strong\u003E\u003Cbr \/\u003E\r\nData-driven paradigms now solve the world\u0026#39;s hardest problems. Instead of authoring a function through exact and explicit rules, machine learning techniques instead learn rules from data.\u0026nbsp;Unfortunately, often these rules are unknown to both the people who train models and the people they impact. This is the purpose and promise of machine learning interpretability. But what is\u0026nbsp;interpretability? What are the rules that modern, complex models learn? And how can we best represent and communicate them to people? Since machine learning now impacts people\u0026#39;s daily lives,\u0026nbsp;we answer these questions taking a\u0026nbsp;\u003Cem\u003Ehuman-centered approach\u003C\/em\u003E\u0026nbsp;by designing and developing interactive interfaces can enable interpretability at scale and for everyone. This thesis focuses on:\u003Cbr \/\u003E\r\n\u003Cbr \/\u003E\r\n\u003Cem\u003E(1) Operationalizing machine learning interpretability:\u003C\/em\u003E We conduct user research for understanding what practitioners want from interpretability, instantiate these into an interactive interface, and use it\u0026nbsp;as a design probe into the emerging practice of interpreting models (Gamut). We also show how different complementary mediums can be used for explanations, such as visualization and\u0026nbsp;verbalization (TeleGam).\u003Cbr \/\u003E\r\n\u003Cbr \/\u003E\r\n\u003Cem\u003E(2) Scaling deep learning interpretability:\u003C\/em\u003E\u0026nbsp;We reveal critical yet missing areas of deep learning interpretability research (Interrogative Survey). For example, neural network explanations focus on single\u0026nbsp;instances or neurons, but predictions are often computed from millions of weights that are optimized over millions of instances, such explanations can easily miss a bigger picture. We present an\u0026nbsp;interactive interface that scalably summarizes and visualizes what features a deep learning model has learned and how those features interact to make predictions (Summit).\u003Cbr \/\u003E\r\n\u003Cbr \/\u003E\r\n\u003Cem\u003E(3) Communicating interpretability with interactive articles:\u003C\/em\u003E\u0026nbsp;Due to rapid democratization, machine learning now impacts everyone. We use interactive articles, a new medium to communicate complex\u0026nbsp;ideas, to communicate interpretability and teach people about machine learning\u0026#39;s capabilities and limitations, while developing new publishing platforms (Parametric Press). We propose a framework for\u0026nbsp;how interactive articles can be effective through a comprehensive survey that pulls from diverse disciplines of research and critical reflections (Interactive Articles).\u003Cbr \/\u003E\r\n\u003Cbr \/\u003E\r\nOur work has made significant impact to industry and society: our visualizations have been deployed and demoed at Microsoft and built into widely-used interpretability\u0026nbsp;toolkits, our research is supported by NASA, and our interactive articles have been read by hundreds of thousands of people.\u003C\/p\u003E\r\n","summary":null,"format":"limited_html"}],"field_subtitle":"","field_summary":"","field_summary_sentence":[{"value":"Interactive Scalable Interfaces for Machine Learning Interpretability"}],"uid":"27707","created_gmt":"2019-11-26 19:36:25","changed_gmt":"2019-11-26 19:36:25","author":"Tatianna Richardson","boilerplate_text":"","field_publication":"","field_article_url":"","field_event_time":{"event_time_start":"2019-12-02T16:30:00-05:00","event_time_end":"2019-12-02T18:30:00-05:00","event_time_end_last":"2019-12-02T18:30:00-05:00","gmt_time_start":"2019-12-02 21:30:00","gmt_time_end":"2019-12-02 23:30:00","gmt_time_end_last":"2019-12-02 23:30:00","rrule":null,"timezone":"America\/New_York"},"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":"78761","name":"Faculty\/Staff"},{"id":"78771","name":"Public"},{"id":"174045","name":"Graduate students"},{"id":"78751","name":"Undergraduate students"}],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[],"email":[],"slides":[],"orientation":[],"userdata":""}}}