Click the links that follow for more news items about Stephen Bach, Zachary Espiritu, Peter Norvig, Roberto Tamassia, and the Randy F. Pausch '82 Computer Science Undergraduate Summer Research Award.
The Randy F. Pausch '82 Computer Science Undergraduate Summer Research Award, given this year to Ross Briden and Zachary Espiritu to support their work with Brown CS Professors Stephen Bach and Roberto Tamassia, respectively, recognizes strong achievement from undergraduate researchers and offers them the opportunity to continue their work over the summer.
A generous gift from Peter Norvig '78 (a Director of Research at Google and a thought leader in the areas of artificial intelligence, natural language processing, information retrieval, and software engineering) established the award, which provides $10,000 annually to support an undergraduate engaged in an intensive faculty-student summer research partnership. The gift honors the life and work of Randy F. Pausch '82, a renowned expert in computer science, human-computer interaction, and design who died of complications from pancreatic cancer in 2008. "His story is inspiring," Peter says, "and this is an opportunity to remember him."
Ross Briden began working with Stephen Bach during the Spring semester of his first year at Brown. "One of Steve’s main research interests," Ross explains, "is weak supervision, which is a technique for labeling data using a variety of noisy data sources. During that time, weak supervision frameworks, such as Snorkel, were really effective when applied to Natural Language Processing (NLP) tasks; however, it wasn’t clear how these techniques could be extended to imagery data. To tackle this problem, I started off by reading papers, reimplementing models, and discussing my ideas with Steve during our weekly meetings. Throughout our discussions, I developed a strong interest in weak supervision as well as machine learning theory. By the spring of my sophomore year, I started to work on TAGLETS, a framework for automatically labeling data using an ensemble of weak labelers, with Top Piriyakulkij, Jeffrey Zhu, and several other students from the BATS lab. The main objective of TAGLETS is to provide a unified framework for labeling data of any modality, generalizing the ideas behind my original research project."
"While working on TAGLETS," Ross says, "I started to explore semi-supervised learning, an approach for training machine learning models using a small amount of labeled data and a large amount of unlabeled data. Semi-supervised learning algorithms make assumptions, such as smoothness, about the distribution that training and test data are drawn from in order label examples. In recent years, semi-supervised learning has exploded in popularity, with many techniques achieving state-of-the-art performance in a variety of domains. However, I noticed that few papers tried to explain why existing semi-supervised learning algorithms were so effective when applied to deep neural networks. My research attempts to address this discrepancy by exploring why, both empirically and theoretically, semi-supervised learning algorithms work at a fundamental level, as well as their limitations. In addition, my research will also explore whether knowledge bases and other forms of supervision can improve the performance of existing semi-supervised learning algorithms."
Ross describes himself as incredibly excited to be able to pursue his research over the summer: "And I look forward to broadening my knowledge of the field. This award will allow me to fully focus on my research and take creative risks I might not otherwise be able to take. I would especially like to thank Steve for his amazing mentorship and support, the BATS lab, and the TAGLETS team."
Searchable encryption (SE), Zachary explains, is a class of privacy-preserving techniques that allow clients to outsource their data to a semi-honest server in an encrypted format. "Recent encrypted database constructions," he says, "allow for practical outsourcing of more expressive types of queries. These expressive types of queries, however, come at a cost to privacy – recent work has shown that realistic kinds of adversaries can infer enough information from observing the 'leakage' revealed by encrypted database queries that allow these adversaries to reconstruct information about the underlying plaintext."
"Much recent work in this area," says Zachary, "focuses on queries as a simple filter operation. However, in real-world applications, users are often not directly interested in the records returned by a query, but rather in the result of a function applied to various attributes of these records. Encrypted databases that support these functions require structures that allow them to serve results efficiently – however, we hypothesize that these efficient representations can result in further leakage. Thus, in this work, we're interested in investigating the potential kinds of attacks that can arise from these fundamental, yet less-investigated kinds of queries."
Zachary's interest in computer security began in the Spring semester of 2018, when he took CSCI 1660 Computer Systems Security with Roberto. "I've since been in close collaboration with Roberto over the last three years," he says, "as I've been fortunate enough to have the opportunity to serve as a Head TA for CSCI 1660 for the last three years in a row. My experience doing course development and design for CSCI 1660 led me to become more interested in privacy-preserving techniques such as multi-party computation and searchable encryption, which ultimately led me to Roberto's area of research in attacks on encrypted databases. I'm really grateful to Roberto for introducing me to computer security in general and encouraging me to get involved in research and on this project. I'm also really grateful to Seny Kamara and my collaborators in the Encrypted Systems Lab, with whom I've been working on related research in the past year on the 'defensive' side of searchable encryption and through whom I've learned a great deal about searchable encryption (and also how to do research in general)."
Ross and Zachary's curiosity and initiative are exactly what Peter Norvig is looking for. He sees this award as a multiplier that will amplify the value of his gift and extend it through time. "In the past," he says, "we had to build all our own tools, and we didn't have time to combine computer science with other fields. Now, there are so many opportunities to do so. I think it's a wise choice: you invest in things that you think will do good, and educating a student allows them to help add to the things that you're already trying to accomplish."
For more information, click the link that follows to contact Brown CS Communication Outreach Specialist Jesse C. Polhemus.