Show simple item record

dc.contributor.advisorDoshi-Velez, Finale
dc.contributor.advisorParkes, David C.
dc.contributor.advisorSinger, Yaron
dc.contributor.authorMasood, Muhammad Arjumand
dc.date.accessioned2019-12-12T09:14:41Z
dc.date.created2019-05
dc.date.issued2019-05-17
dc.date.submitted2019
dc.identifier.citationMasood, Muhammad Arjumand. 2019. Algorithms for Discovering Collections of High-Quality and Diverse Solutions, With Applications to Bayesian Non-Negative Matrix Factorization and Reinforcement Learning. Doctoral dissertation, Harvard University, Graduate School of Arts & Sciences.
dc.identifier.urihttp://nrs.harvard.edu/urn-3:HUL.InstRepos:42029756*
dc.description.abstractMachine Learning problems often admit a solution space that is not unique. When multiple feasible solutions exist, picking from a diverse, representative set may lead to better generalization and task-specific performance. While the emphasis of much of the literature has been on directly finding the `best' solution, we show that often a diverse set of near optimal solutions can be found which may be useful to practitioners and experts using machine learning models in decision making. This thesis investigates methods for obtaining a useful collection of solutions in specific models. Non-negative Matrix Factorization (NMF) is a popular data exploration tool and its Bayesian formulation is a promising approach for understanding uncertainty within this structure. We demonstrate that current approaches are lacking in the proper characterization of uncertainties and present novel techniques to provide model flexibility and improve the quality and speed of the inference. These techniques are applied to standard benchmark datasets for NMF as well as a curated medical dataset for understanding comorbidities in the Autism Spectrum Disorder (ASD). We show how a distinct collection of NMFs of nearly equal quality give rise to variability in interpretation of features and subsequent predictions. Finally, we present extensions of our diverse collection-based approach to the on-policy and off-policy Reinforcement Learning setting. Here, a completely new set of technical tools is required. In both on-policy and off-policy variants, we use diversity as a regularization feature in order to obtain a set of high-quality diverse policies. In addition to finding diverse policies in simulate-able multi-goal domains, we find a diverse set of policies designed to aid clinical decision making using ICU data for sepsis and hypotension management.
dc.description.sponsorshipEngineering and Applied Sciences - Applied Math
dc.format.mimetypeapplication/pdf
dc.language.isoen
dash.licenseLAA
dc.subjectmachine learning
dc.subjectNMF
dc.subjectnon-negative matrix factorization
dc.subjectreinforcement learning
dc.subjectpolicy gradient
dc.subjectStein discrepancy
dc.titleAlgorithms for Discovering Collections of High-Quality and Diverse Solutions, With Applications to Bayesian Non-Negative Matrix Factorization and Reinforcement Learning
dc.typeThesis or Dissertation
dash.depositing.authorMasood, Muhammad Arjumand
dc.date.available2019-12-12T09:14:41Z
thesis.degree.date2019
thesis.degree.grantorGraduate School of Arts & Sciences
thesis.degree.grantorGraduate School of Arts & Sciences
thesis.degree.levelDoctoral
thesis.degree.levelDoctoral
thesis.degree.nameDoctor of Philosophy
thesis.degree.nameDoctor of Philosophy
dc.type.materialtext
thesis.degree.departmentEngineering and Applied Sciences - Applied Math
thesis.degree.departmentEngineering and Applied Sciences - Applied Math
dash.identifier.vireo
dc.identifier.orcid0000-0002-9494-8307
dash.author.emailarjumand.masood@gmail.com


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record