Osama Hanna

Biography

Osama Hanna is a Research Scientist at Meta, AI, and Ph.D. candidate at the Electrical and Computer Engineering Department at UCLA. He works with Prof. Christina Fragouli (advisor) and Prof. Lin F. Yang (co-advisor) as a member of ARNI lab, and he closely collaborates with Prof. Suhas Diggavi. Before joining UCLA, he received his Bachelor’s and Masters degrees in Electrical, Electronics and Communications Engineering from Cairo University, and Nile University respectively.

Research Interests

• Bandits and reinforcement learning
• Learning theory
• Information and coding theory
• Algorithms
• Mathematical logic

My research explores the intersection of online learning and information theory. Machine learning techniques guide in devising better algorithms, while information theory sets fundamental barriers for optimal performance. However, things happened to work in reverse for me:) my struggles with machine learning techniques to enhance performance identified the challenging aspects, hence, helped me to establish performance lower bounds, while struggling with information-theoretic lower bounds led to improved algorithms. I enjoy the process of thinking about the mathematical and modeling aspects of these problems, and hope my research will advance interaction between humans and AI, enhancing decision-making in areas such as healthcare, inclusive prosperity, and economic development.

Education

• Ph.D., Electrical and Computer Engineering, University of California Los Angeles (Advisors: Prof. Christina Fragouli, and Lin F. Yang)
  Although I am with the ECE department you were most likely to find me in Math department halls. Why? I completed 7 ECE courses, but my love affair with Math reached 12 courses.
  Collaborators at UCLA: Christina Fragouli (advisor), Lin F. Yang (co-advisor), and Suhas Diggavi
• M.Sc., Electrical Engineering, Nile University, Egypt
• B.S., Electrical, Electronics and Communications Engineering, Cairo University, Egypt

Experience

• Reserach Scientist, Meta AI (2024-current)
• Research Intern, Meta AI (Summer 2022)
• Research Intern, Aalborg University, Denmark (Advisor: Petar Popovski) (Summer 2017)

Publications

Journal Papers

• Osama Hanna, Lin F. Yang, and Christina Fragouli, “Contexts can be Cheap: A Reduction from Stochastic Contextual Bandits to Linear Bandits,” under submission.
• Osama Hanna, Xinlin li, Christina Fragouli, and Suhas Diggavi, “Common information dimension,” under submission.
• Osama Hanna, Merve Karakas, Lin F. Yang, and Christina Fragouli, “Multi-Arm Bandits over Action Erasure Channels,” under submission.
• Osama A. Hanna, Antonious M. Girgis, Christina Fragouli, and Suhas Diggavi, “Differentially Private Stochastic Linear Bandits:(Almost) for Free,” (JSAIT), 2024.
• Osama A. Hanna, Lin F. Yang, and Christina Fragouli, “Compression for Multi-Arm Bandits,” in Journal on Selected Areas in Information Theory (JSAIT), 2023.
• Osama A. Hanna, Yahya H. Ezzeldin, Christina Fragouli, and Suhas Diggavi, “Quantization of Distributed Data for Learning”, in Journal on Selected Areas in Information Theory (JSAIT), 2021.
• Osama A. Hanna, Yahya H. Ezzeldin, T. Sadjadpour, Christina Fragouli and Suhas Diggavi, “On Distributed Quantization for Classification”, in IEEE Journal on Selected Areas in Information Theory 2020.
• Osama A. Hanna, M. Nafie, and A. El-Keyi, “Cache-Aware Source Coding”, in IEEE Comm. Letters, 2018.

Conference Papers

• Osama Hanna, Merve Karakas, Lin F. Yang, and Christina Fragouli, “ulti-Agent Bandit Learning through
  Heterogeneous Action Erasure Channels,” in AISTATS 2024.
• Osama A. Hanna, Lin F. Yang, and Christina Fragouli, “Efficient Batched Algorithm for Contextual Linear Bandits with Large Action Space via Soft Elimination,” in NeurIPS 2023.
• Osama A. Hanna, Lin F. Yang, and Christina Fragouli, “Contexts can be Cheap: Solving Stochastic Contextual Bandits with Linear Bandit Algorithms,” in Conference on Learning Theory (COLT) 2023.
• Osama A. Hanna, Xinlin li, Christina Fragouli, and Suhas Diggavi, “Common information dimension,” in ISIT 2023.
• Osama A. Hanna, Merve Karakas, Lin F. Yang, and Christina Fragouli, “Multi-Arm Bandits over Action Erasure Channels,” in ISIT 2023.
• Xinlin li, Osama Hanna, Christina Fragouli, Suhas Diggavi, Gunjan Verma, and Joydeep Bhattacharyya, “Feature Compression for Multimodal Multi-Object Tracking,” in MILCOM 2023.
• Xinlin li, Merve Karakas, Osama Hanna, Mehrdad Kiamari, Jared Coleman, Christina Fragouli, Bhaskar Krishnamachari, and Gunjan Verma, “Online Allocation of Sensing and Computation,” in IEEE CIC 2023.
• Osama A. Hanna, Lin F. Yang, and Christina Fragouli, “Learning from Distributed Users in Contextual Linear Bandits Without Sharing the Context”, in NeurIPS 2022.
• Osama A. Hanna, Lin F. Yang, and Christina Fragouli, “Solving Multi-Arm Bandit Using a Few Bits of Communication”, in AISTATS 2022.
• Osama A. Hanna, Xinlin li, Christina Fragouli, and Suhas Diggavi, “Can we break the dependency in distributed detection?,” in ISIT 2022.
• Osama A. Hanna, Lin F. Yang, and Christina Fragouli, “Solving Multi-Arm Bandit Using a Few Bits of Communication”, spotlight talk in ICML Workshop on Reinforcement Learning Theory 2021.
• Rasmus Vestergaard, Osama A. Hanna, Linqi Song, Daniel E. Lucani, and Christina Fragouli, “On Coded Broadcasting for Wireless Recommendation Systems”, in IEEE ICC, 2021.
• Anders E. Kalor, Osama A. Hanna, and Petar Popovski, “Random access schemes in wireless systems with correlated user activity”, in IEEE 19th Int. Workshop on Signal Proc. Advances in Wireless Comm. (SPAWC), 2018.
• Osama A. Hanna, Amr El-Keyi, and Mohammed Nafie, “Degrees of Freedom in Cached MIMO Relay Networks With Multiple Base Stations,” in International Wireless Communications and Mobile Computing Conference (IWCMC).

Invited Talks

• [April 2024] University of California, Riverside
• [April 2024] Stanford University
• [Feb 2024] University of California, Berkeley
• [Jan 2024] University of Southern California
• [Nov 2023] The Reinforcement Learning Theory Seminar Series
• [Nov 2023] University of California, Irvine
• [Jul 2023] The Chinese University of Hong Kong
• [Jul 2023] City University of Hong Kong
• [Feb 2023] ITA Graduation Day Talk
• [Jan 2022] Spotlight talk at ICML Workshop on Reinforcement Learning Theory

Teaching

• [Fall 2019] Teaching Assistant for Linear Programming (EE236A), University of California, Los Angeles
• [Fall 2021] Teaching Assistant for Graph Theory (EE134), University of California, Los Angeles

Mentoring

• PhD students: Merve Karakas, Xinlin Li
• Undergraduate students: Tara Sadjadpour, Dylan Lee, Yiming Guo, Karim Saraipour, Zixiang Ji, Jianxing
  Zhao

Reviewing

NeurIPS, ICML, ALT, ICLR, AISTATS, ISIT, GLOBECOM,  IEEE JSAIT, IEEE JSAC, IEEE Comm Letters, IEEE TWC, IEEE TC

Selected Courses from ECE Department at UCLA

Reinforcement Learning Theory(A+), Information Theory(A+), Foundations of Statistical Learning(A), Linear Programming(A), Convex Optimization(A), Large Scale Optimization(A)

Selected Courses from Math Department at UCLA

Algorithms(A+), Computability Theory(A+), Mathematical Logic(A+), Axiomatic Set Theory(A), Analysis Honors(A+), Topology(A), Optimization(A+), Model Theory(A), Measure Theory(A), Calculus of Variations(A)