79 Amazon Research Awards recipients announced

Awardees, who represent 54 universities in 14 countries, have access to Amazon public datasets, along with AWS AI/ML services and tools.

By Amazon Research Awards team

April 17, 2023

8 min read

Amazon Research Awards (ARA) provides unrestricted funds and AWS Promotional Credits to academic researchers investigating various research topics in multiple disciplines. This cycle, ARA received many excellent research proposals from across the world and today is publicly announcing 79 award recipients who represent 54 universities in 14 countries.

This announcement includes awards funded under four call for proposals during the fall 2022 cycle: AWS AI, Automated Reasoning, Prime Video, and Sustainability. Proposals were reviewed for the quality of their scientific content and their potential to impact both the research community and society.

Recipients have access to more than 300 Amazon public datasets and can utilize AWS AI/ML services and tools through their AWS Promotional Credits. Recipients also are assigned an Amazon research contact who offers consultation and advice, along with opportunities to participate in Amazon events and training sessions.

Additionally, Amazon encourages the publication of research results, presentations of research at Amazon offices worldwide, and the release of related code under open-source licenses.

"Complexities of AI/ML challenges at scale often intersect more than one discipline and require creative and diverse approaches to tackle these issues," said Arash Nourian, AWS general manager, Machine Learning Engines. "I was amazed by the diversity of disciplines and the scientific content of Awardee’s submissions that collectively could represent significant potential impact on both the AI/ML research community and society."

“The incredible response to Prime Video’s fall 2022 Call for Proposals is a testament to the exciting work the ARAs have inspired across four cutting-edge research categories,” said BA Winston, VP of Technology at Prime Video. “I am delighted by the winning proposals and look forward to the ongoing research across several areas in Prime Video that is helping us create even more impactful customer-obsessed technology.”

ARA funds proposals throughout the year in a variety of research areas. Applicants are encouraged to visit the ARA call for proposals page for more information or send an email to be notified of future open calls.

The tables below list, in alphabetical order, fall 2022 cycle call-for-proposal recipients, sorted by research area.

AWS AI

Recipient	University	Research title
Jonathan Afilalo	McGill University	Coreslicer: deep learning of CT images for frailty assessment in clinical care
Saman Amarasinghe	Massachusetts Institute of Technology	Reimagining the compiler in the cloud
Akshay Chaudhari	Stanford University	Large-scale self-supervised learning for medical imaging
Soheil Feizi	University Of Maryland, College Park	Towards mitigating spurious correlations in deep learning
Aikaterini Fragkiadaki	Carnegie Mellon University	Analogical networks for continual memory-modulated visual learning and language understanding
Mark Gerstein	Yale University	Privacy-preserving storage, sharing, and analysis for genomics data
Joseph Gonzalez	University Of California, Berkeley	A unified platform for training and serving large models
Michael Gubanov	Florida State University	An interactive polygraph for robust access to scientific knowledge
Yan Huang	Carnegie Mellon University	Combating algorithmic bias inherited from human decision making: a human-AI perspective
CV Jawahar	The International Institute of Information Technology - Hyderabad	Deeper understanding of multilingual handwritten documents: from recognition to dialogues
Zhihao Jia	Carnegie Mellon University	Combining ML and systems optimizations for sustainable and affordable ML
Daniel Khashabi	Johns Hopkins University	Crowdsourcing with machine backbone
Rahul Krishnan	University Of Toronto	Towards a learning healthcare system
Anastasios Kyrillidis	Rice University	Efficient and affordable transformers for distributed platforms
Kevin Leach	Vanderbilt University	Documentnet: iterative data collection for building a robust document understanding dataset
Lei Li	University Of California, Santa Barbara	Real-time robust simultaneous interpretation with few samples
Xiaoyi Lu	University Of California, Merced	Scaling collective communication for distributed deep learning training
Yunan Luo	Georgia Institute of Technology	Calibrated and interpretable geometric deep learning for robust drug screening
Graham Neubig	Carnegie Mellon University	Towards more reliable and interpretable code language models
Qing Qu	University of Michigan, Ann Arbor	Principles of deep representation learning via neural collapse
Mirco Ravanelli	Concordia University	Toward empathetic conversational AI
Amit Roy-Chowdhury	University of California, Riverside	Exploring privacy in deep metric learning: applications in computer vision
Chirag Shah	University of Washington	Fairness as a service: operationalizing fairness in search and recommendation applications through a novel multi-objective optimization framework
Kristina Simonyan	Massachusetts Eye and Ear/Harvard Medical School	Machine learning for automated speech processing for real-time speech prosthesis in neurological disorders
Berrak Sisman	University of Texas, Dallas	Explainable AI for expressive voice synthesis
Dawn Song	University Of California, Berkeley	FedOps: an abstraction for trustworthy federated learning
Peter Spirtes	Carnegie Mellon University	System-level and long-term fairness through causal learning and reasoning
Ion Stoica	University Of California, Berkeley	A unified platform for training and serving large models
Vasileios Syrgkanis	Stanford University	Automating the causal machine learning pipeline
Carlo Tomasi	Duke University	Deep neural network classifiers with margins in input space
Yatish Turakhia	University Of California, San Diego	Machine learning enabled wastewater-based epidemiology
Xiaolong Wang	University of California, San Diego	Learning implicit neural foundation models
Neeraja Yadwadkar	University Of Texas, Austin	Easy-to-use and cost-efficient distributed inference serving
Hamed Zamani	University Of Massachusetts Amherst	On the optimization of retrieval-enhanced machine learning models
Ce Zhang	ETH Zurich	FedOps: an abstraction for trustworthy federated learning
Tianyi Zhang	Purdue University	Human-in-the-loop deep learning optimization for better usability, transparency, and user trust
Yiying Zhang	University Of California, San Diego	Training deep neural networks with “zero” activations
Jishen Zhao	University Of California, San Diego	Semantic-informed document structure recognition with large language models
Ben Zhao	University Of Chicago	Digital forensics for deep neural networks
Heather Zheng	University of Chicago	Digital forensics for deep neural networks
Jun-Yan Zhu	Carnegie Mellon University	Compositional personalization of large-scale generative models
Jia Zou	Arizona State University	A compilation framework for accelerating machine learning inference queries

Amazon Sustainability

Recipient	University	Research title
Vikram Iyer	University of Washington	Computational design and circular fabrication for sustainable electronics
Adriana Schulz	University of Washington	Computational design and circular fabrication for sustainable electronics
Mari Winkler	University of Washington	A novel bioreactor platform for continuous high‐rate bio-production

Automated Reasoning

Recipient	University	Research title
Maria Paola Bonacina	Università degli Studi di Verona	Advances in conflict-driven SATisfiability modulo theories and assignments
Ahmed Bouajjani	Universite Paris-Cite	Safe composition of distributed off-the-shelf components
Martin Nyx Brain	City, University Of London	Snowshoes: overapproximating code footprints for safe program exploration
Anton Burtsev	University Of Utah	Atmosphere: leveraging language safety and operating system design for verification
Alastair Donaldson	Imperial College London	DafnyDefender: automated testing for the Dafny ecosystem
Francois Dupressoir	University Of Bristol	Formosa cryptography: computer-aided reasoning for high-assurance cryptographic design and engineering
Gidon Ernst	Ludwig Maximilian University of Munich	Security specifications for Dafny
Pascal Fontaine	University of Liège	SMT: modules, formats, and standards
Jeffrey Foster	Tufts University	Automated testing of external methods in Dafny
Sicun Gao	University Of California, San Diego	Monte Carlo tree methods for decision-making in dReal
Philippa Gardner	Imperial College London	Gillian-Rust: unbounded verification for unsafe rust code
Limin Jia	Carnegie Mellon University	Enabling one-line rust verification with program synthesis
Patrick Lam	University Of Waterloo	Statically inferring contracts from assertions & tests
Aravind Machiry	Purdue University	Security verification and hardening of CI workflows
Anders Møller	Aarhus University	Securing node.js programs with static resource analysis
Magnus Myreen	Chalmers University Of Technology	Compiling Dafny to CakeML
ThanhVu Nguyen	George Mason University	Scalable and precise DNN constraint solving with abstraction and conflict clause learning
Burcu Kulahcioglu Ozkan	Delft University of Technology	Coverage-directed randomized testing of distributed systems
Bryan Parno	Carnegie Mellon University	Verus: developing provably correct and reliable rust code
Corina Pasareanu	Carnegie Mellon University	Enabling one-line rust verification with program synthesis
Ruzica Piskac	Yale University	Formalizing FISA: using automated reasoning to formalize legal reasoning
Elizabeth Polgreen	University of Edinburgh	Automated and provably correct code modernization
Fred Schneider	Cornell University	Using non-deterministic executable specification to test properties that relate executions
Scott Shapiro	Yale University	Formalizing FISA: using automated reasoning to formalize legal reasoning
Marc Shapiro	INRIA & Sorbonne Universite Paris	Safe composition of distributed off-the-shelf components
Alexandra Silva	Cornell University	Automated reasoning for correctness and incorrectness
Yakir Vizel	Technion – Israel Institute Of Technology	Lazy and incremental framework for solving CHCs
Florian Zuleger	Technische Universität Wien	Automated cost analysis of data structures

Prime Video

Recipient	University	Research title
David Bull	University of Bristol	Generic deep video quality assessment in the extended parameter space
Eamonn Keogh	University of California Riverside	A proposal to make any time series anomaly detection algorithm faster, more accurate and more practical
Xiaorui Liu	North Carolina State University	Deep reinforcement learning for the mixed ranking of recommendations and advertisements with page-wise display
Jiliang Tang	Michigan State University	Deep reinforcement learning for the mixed ranking of recommendations and advertisements with page-wise display
Hanghang Tong	University of Illinois Urbana-Champaign	Graph algorithms for personalized recommendation
Fan Zhang	University of Bristol	Generic deep video quality assessment in the extended parameter space