Events

Hide Filters Show Filters
Hide Filters Show Filters

Event Type

Impact Area

Research Area

Seminar Series

Reset Filters

February 01, 2023

No events scheduled

February 02, 2023

Private Event

Information session for Beyond CSAIL

3:00P
- 5:00P

Location

32-G449
Patil/Kiva Seminar Room
Add to Calendar 2023-02-02 15:00:00 2023-02-02 17:00:00 America/New_York Information session for Beyond CSAIL Learn about a new startup support program for the CSAIL community, launching on February 2, 2023. The program is called Beyond CSAIL and is offered in collaboration with E14 Fund.Beyond CSAIL is a set of programs and resources available for MIT students and postdocs thinking about starting a company. The available programs include travel grants, fellowship investments, access to mentorship, or resources. The inaugural (annual) launch event will be held on February 2, 2023 in Patil/Kiva Seminar Room 32-G440, in Stata Center/Bldg 32. Join us for the launch of Beyond CSAIL, where you'll have the opportunity to tap into a wealth of resources and support as you embark on your entrepreneurial journey, from E14 Fellowship Investments to mentorship from industry experts and VCs. Hear from successful CSAIL startup founders about their journey, and then connect with the community over food and drinks. Let's make your entrepreneurial dream a reality together. Following the event, there will be a mixer off campus at 5:30pm*Register to join us at https://btlcsail.eventbrite.com/*Agenda:Talk and Panel at Patil/Kiva Seminar Room, 32-G4493:00 - 3:45pm: Welcome and Program Overview (Daniela Rus, Habib Haddad, Calvin Chin)3:45- 4:45pm: CSAIL startup panel (Tutor Intelligence, Secure AI Labs, Themis AI)4:45-5:00pm: Q&AVenture Mixer at CommonWealth: 1 Broad Canal Way, Kendall Square, Cambridge, MA 021425:30pm-8:30pm: Food and drinksMore info at https://btl.mit.edu/csail 32-G449

February 03, 2023

February 08, 2023

A World Wide View of the World Wide Web

Kimberly Ruth
Stanford University
4:00P
- 6:00P
Add to Calendar 2023-02-08 16:00:00 2023-02-08 18:00:00 America/New_York A World Wide View of the World Wide Web Despite the pervasiveness of the modern web, we know surprisingly little about how people use it. In this talk, I’ll discuss two recent works aimed at understanding what web browsing looks like today, both appearing at IMC 2022. First, I’ll analyze lists of most popular websites, such as the Alexa Top Million, that are widely used in security and networking research but are not well vetted for accuracy. I’ll show that most top website lists capture web popularity poorly, with the exception of the Chrome User Experience Report dataset. Second, I’ll describe the first large-scale study of how people spend time on the web, using data from Chrome browser telemetry spanning hundreds of millions of users. I’ll explore how traffic is distributed among top websites, where people spend time on the web, and what differences we see in web browsing geographically. Finally, I’ll conclude with recommendations for how the research community can more accurately analyze the web in the future.Join Zoom Meetinghttps://mit.zoom.us/j/97527284254Password: <3securityOne tap mobile+16465588656,,97527284254# US (New York)+16699006833,,97527284254# US (San Jose)Meeting ID: 975 2728 4254US : +1 646 558 8656 or +1 669 900 6833International Numbers: https://mit.zoom.us/u/auBvg4NEVJoin by SIP97527284254@zoomcrc.comJoin by Skype for Businesshttps://mit.zoom.us/skype/97527284254

February 10, 2023

Nearly Optimal Property Preserving Hashing

LaKyah Tyner
Northeastern University
10:30A
- 12:00P

Location

G-882, Hewlett Room
Add to Calendar 2023-02-10 10:30:00 2023-02-10 12:00:00 America/New_York Nearly Optimal Property Preserving Hashing Property-preserving hashing (PPH) consists of a family of compressing hash functions h such that, for any two inputs x, y, we can correctly identify whether some property P(x, y) holds given only the digests h(x), h(y). In a basic PPH, correctness should hold with overwhelming probability over the choice of h when x, y are worst-case values chosen a-priori and independently of h. In an adversarially robust PPH (RPPH), correctness must hold even when x, y are chosen adversarially and adaptively depending on h. Here, we study (R)PPH for the property that the Hamming distance between x and y is at most t.       The notion of (R)PPH was introduced by Boyle, LaVigne and Vaikuntanathan (ITCS ’19), and further studied by Fleischhacker, Simkin (Eurocrypt ’21) and Fleischhacker, Larsen, Simkin (Eurocrypt ’22). In this work, we obtain improved constructions that are conceptually simpler, have nearly optimal parameters, and rely on more general assumptions than prior works. Our results are: We construct information-theoretic non-robust PPH for Hamming distance via syndrome listdecoding of linear error-correcting codes. We provide a lower bound showing that this construction is essentially optimal.We make the above construction robust with little additional overhead, by relying on homomorphic collision-resistant hash functions, which can be constructed from either the discrete-logarithm or the short-integer-solution assumptions. The resulting RPPH achieves improved compression compared to prior constructions, and is nearly optimal.We also show an alternate construction of RPPH for Hamming distance under the minimal assumption that standard collision-resistant hash functions exist. The compression is slightly worse than our optimized construction using homomorphic collision-resistance, but essentially matches the prior state of the art constructions from specific algebraic assumptions.Lastly, we study a new notion of randomized robust PPH (R2P2H) for Hamming distance, which relaxes RPPH by allowing the hashing algorithm itself to be randomized. We give an informationtheoretic construction with optimal parameterSpeaker Bio: LaKyah is a second-year student at Northeastern University working with Daniel Wichs and abhi shelat. She is currently doing work with     property preserving hashing and threshold signatures. G-882, Hewlett Room

February 14, 2023

Pedro Lopes - Integrating interactive devices with the user’s body

Pedro Lopes
4:00P
- 5:00P

Location

32-D463 (Star)
Add to Calendar 2023-02-14 16:00:00 2023-02-14 17:00:00 America/New_York Pedro Lopes - Integrating interactive devices with the user’s body Abstract:The main question that drives my research is: what is the next interface paradigm that supersedes wearable devices? I argue that the new paradigm is one in which interactive devices will integrate with the user’s biological senses and actuators.This way of engineering devices that intentionally borrow parts of the user’s biology puts forward a new generation of miniaturized devices; allowing us to circumvent traditional physical constraints. For instance, in the case of my devices based on electrical muscle stimulation, they demonstrate how our body-device integration circumvents the constraints imposed by the size of motors used in traditional haptic devices (e.g., robotic exoskeletons). Taking this further, we can apply this integrated approach to other modalities. For instance, we engineered a device that delivers chemicals to the user to generate temperature sensations, without the need to rely on cumbersome thermal actuators, such as air conditioners or heaters. My approach to miniaturizing devices is especially useful to advance mobile interactions, such as in virtual or augmented reality, where users have a desire to remain untethered & free.Integrating devices with the user’s body allows to give users new physical abilities. For example, we have engineered a device that allows users to locate odor sources by “smelling in stereo” as well as a device that physically accelerates the user’s reaction time using muscle stimulation, which allows users to steer to safety or even catch a falling object that they would normally miss.While this integration can offer many benefits (e.g., faster reaction time, realistic simulations in VR/AR, or faster skill acquisition), it also requires tackling new challenges, such as the question of agency: do we feel in control when our body is integrated with an interface? Together with our colleagues in neuroscience, we have been measuring how our brain encodes agency to improve the design of this new type of integrated interfaces. We found that, even in the extreme case of interfaces that electrically control the user’s muscles, it is possible to improve the sense of agency. More importantly, we found that it is only by preserving the user’s sense of agency that these integrated devices provide benefits even after the user takes them out. Bio:Pedro Lopes is an Assistant Professor in Computer Science at the University of Chicago. Pedro focuses on integrating computer interfaces with the human body—exploring the interface paradigm that supersedes wearable computing. Some of these new integrated-devices include: muscle stimulation wearable that allows users to manipulate tools they have never seen before or that accelerate their reaction time, or a device that leverages the sense of smell to create an illusion of temperature. Pedro’s work has received a number of academic awards, such as five ACM CHI/UIST Best Papers, Sloan Fellowship and NSF CAREER. Pedro’s research also captured the interest of the public & media, covered by the New York Times and was exhibited at Ars Electronica & World Economic Forum.The talk will also be streamed over Zoom: https://mit.zoom.us/j/96214841975. 32-D463 (Star)

February 15, 2023

Generative Machines and Ground Truth

Kate Crawford
Microsoft Research

Part Of

5:00P
- 6:00P

Location

32-123
Kirsch Auditorium
Add to Calendar 2023-02-15 17:00:00 2023-02-15 18:00:00 America/New_York Generative Machines and Ground Truth Abstract:We are living in a period of rapid acceleration for generative AI, where large language and text-to-image diffusion models are being deployed in a multitude of everyday contexts. From ChatGPT’s training set of hundreds of billions of words to LAION-5B’s corpus of almost 6 billion image-text pairs, these vast datasets – scraped from the internet and treated as “ground truth” – play a critical role in shaping the epistemic boundaries that govern machine learning models. Yet training data is beset with complex social, political, and epistemological challenges. What happens when data is stripped of context, meaning, and provenance? How does training data limit *what* and *how* machine learning systems interpret the world? And most importantly, what forms of power do these approaches enhance and enable? This lecture is an invitation to reflect on the epistemic foundations of generative AI, and to consider the wide-ranging impacts of the current generative turn. Bio:Professor Kate Crawford is a leading international scholar of the social implications of artificial intelligence. She is a Research Professor at USC Annenberg in Los Angeles, a Senior Principal Researcher at MSR in New York, an Honorary Professor at the University of Sydney, and the inaugural Visiting Chair for AI and Justice at the École Normale Supérieure in Paris. Her latest book, Atlas of AI (Yale, 2021) won the Sally Hacker Prize from the Society for the History of Technology, the ASSI&T Best Information Science Book Award, and was named one of the best books in 2021 by New Scientist and the Financial Times. Over her twenty-year research career, she has also produced groundbreaking creative collaborations and visual investigations. Her project Anatomy of an AI System with Vladan Joler is in the permanent collection of the Museum of Modern Art in New York and the V&A in London, and was awarded with the Design of the Year Award in 2019 and included in the Design of the Decades by the Design Museum of London. Her collaboration with the artist Trevor Paglen, Excavating AI, won the Ayrton Prize from the British Society for the History of Science. She has advised policy makers in the United Nations, the White House, and the European Parliament, and she currently leads the Knowing Machines Project, an international research collaboration that investigates the foundations of machine learning. 32-123

February 17, 2023

Functional Commitments for All Functions, under Transparent Setup and Standard Lattice Assumptions

Leo de Castro
10:30A
- 12:00P

Location

RM G-882
Add to Calendar 2023-02-17 10:30:00 2023-02-17 12:00:00 America/New_York Functional Commitments for All Functions, under Transparent Setup and Standard Lattice Assumptions A functional commitment scheme enables a user to concisely commit to a function from a specified family, then later concisely and verifiably reveal values of the function at desired inputs. Useful special cases, which have seen applications across cryptography, include vector commitments and polynomial commitments. To date, functional commitments have been constructed (under falsifiable assumptions) only for functions that are essentially linear.In this talk, we give the first functional commitment scheme for nonlinear functions — indeed, for all functions of any bounded complexity — under a standard setup and a falsifiable assumption. More specifically, the setup is "transparent," requiring only public randomness (and not any trusted entity), and the assumption is the hardness of the standard Short Integer Solution (SIS) lattice problem. Our construction also has other attractive features, including: stateless updates via generic composability; excellent asymptotic efficiency for the verifier, and also for the committer in important special cases like vector and polynomial commitments, thanks to preprocessing (which can even be outsourced to an untrusted party); and post-quantum security, since it is based on SIS. This is joint work with Chris Peikert. RM G-882

March 22, 2023