Events

Hide Sidebar Show Sidebar
Hide Sidebar Show Sidebar
<%= month %>, <%= year %>
<%= moment().add(1, 'month').format('MMM') %>
<% for(var i = 0; i < daysOfTheWeek.length; i++) { %> <% } %> <% for(var i = 0; i < numberOfRows; i++){ %> <% for(var j = 0; j < 7; j++){ %> <% var d = j + i * 7; %> <% } %> <% } %>
<%= daysOfTheWeek[i] %>
<%= days[d].day %>

Current Seminar Series

June 10, 2025

Visipedia 2025 @ MIT

9:00A
- 3:00P

Location

32-G449
Patil/Kiva
Add to Calendar 2025-06-10 9:00:00 2025-06-10 15:00:00 America/New_York Visipedia 2025 @ MIT Visipedia is a long-standing research consortium envisioned back in 2008 by Serge Belongie and Pietro Perona that seeks to develop interactive and participatory systems that capture and share visual expertise. These types of human- or expert-in-the-loop systems have proven to be wildly successful when deploying AI in real-world, expert scenarios from ecology to medicine. Today, AI systems look much different than they did a decade ago, and the ways that we interact with AI have diversified. What does “human-in-the-loop” mean for the next 10 years? Join the Visipedia team on June 10th at MIT for a morning of discussion and brainstorming on the future of expertise and its intersection with AI. Speakers:Bailey Flanigan (Harvard)Vincent Sitzmann (MIT)David Fouhey (NYU)Mark Hamilton (MIT)Antonio Torralba (MIT)Michael Hobley (Caltech)Justin Kay (MIT)Max Hamilton (UMass Amherst)Sara Beery (MIT) The event is open to the public, please feel free to share with your colleagues. We look forward to seeing you! TBD

June 12, 2025

From the Front Lines to the Test Bench: Applied AI and RAG Evaluation, ft. BCG

12:00P
- 1:00P

Location

32-D463
Star
Add to Calendar 2025-06-12 12:00:00 2025-06-12 13:00:00 America/New_York From the Front Lines to the Test Bench: Applied AI and RAG Evaluation, ft. BCG BCG X is thrilled to host an exclusive in-person Tech Talk for the MIT CSAIL community featuring Harnish Jani (https://www.linkedin.com/in/harnishjani/), Managing Director and Partner at BCG X, and Max Struever (https://www.linkedin.com/in/maxstruever/), Principal at BCG X. Join us for two dynamic discussions exploring GenAI-augmented field operations and retrieval-augmented generation (RAG) evaluation. Discussion A (12:00 PM – 12:30 PM ET) Session Title: Accelerating Value Unlock for Field Operations, Maintenance, and Safety Through (Gen)AI Led by: Harnish Jani In high-stakes, hands-on environments like field services, the potential for AI is massive—but realizing that potential isn’t straightforward. The explosive emergence of Generative AI has dominated headlines, leading innovators and operators in the field services sector to ask deeper questions: What does this technology actually mean for operations on the ground? How do we drive adoption among technicians? And how can we turn hype into tangible impact? In this session, we’ll explore practical applications of (Gen)AI across field operations, maintenance, and safety—grounded in lessons from real-world deployments. We'll highlight the roles being transformed and how to build trust for creating next-gen field technicians. Key topics include: Artificial Intelligence, Generative AI Next-Gen Field Technicians (Upskilling and Workforce Transformation) Trust and Adoption in (Gen)AI Implementation Discussion B (12:30 PM – 1:00 PM ET) Session Title: Do You Know What You’re Testing? A Generalizable Coverage Metric for RAG Evaluation. Led by: Max Struever While most LLM evaluation tools focus on output quality, they often overlook a critical blind spot: whether the evaluation questions themselves comprehensively and representatively test the underlying knowledge base. In this session, we’ll introduce a new, generalizable methodology for assessing test set coverage in retrieval-augmented generation (RAG) systems. This approach is especially important in constrained domains like enterprise RAG applications. We’ll present a metric-based framework using vector embeddings to quantify coverage, share real-world examples where coverage diagnostics enhanced evaluation validity, and explore broader implications for benchmarking, system tuning, and error analysis. Key topics include: RAG: Retrieval-Augmented Generation LLM Evaluation Coverage Metrics  We hope you’ll join us for this engaging hour of insights and conversation. Lunch will be served, and there will be time for live Q&A with both speakers. Reserve your seat now and be part of the conversation shaping the future of tech! REGISTRATION: https://airtable.com/appbkKhr4y8cVCCdm/shr8GUIWjRQxIhjsv  TBD

June 16, 2025

A renewed focus on the structure of concurrent data structures

Mike Spear
Lehigh University
4:00P
- 5:00P

Location

Zoom
Virtual
Add to Calendar 2025-06-16 16:00:00 2025-06-16 17:00:00 America/New_York A renewed focus on the structure of concurrent data structures Abstract:It is common for concurrent data structure designers to treat the limitations of their chosen synchronization mechanism as the most important design characteristic. This can overly constrain programmers, especially if too much emphasis is placed on employing a single linearizing compare-and-swap.In this talk, I will discuss recent innovations in synchronization, and then argue that they enable programmers to place their focus back where it belongs: on the structure of concurrent data structures. I will describe exoTM, one (of many!) recent synchronization frameworks that blurs the line between transactional memory and multi-word compare-and-swap (MCAS). ExoTM lets programmers think of synchronization in terms of coarse-grained transactions, while retaining the ability for low-level optimization. I will then describe the Skip Hash, a shockingly simple and easy-to-verify concurrent ordered map that uses exoTM's composability and novel atomicity features to reduce many operations' complexity to O(1).---Bio:Michael Spear is an associate professor in the Computer Science and Engineering Department at Lehigh University. He received a BS from the United States Military Academy and a PhD from the University of Rochester. His research focuses on issues related to scalability and programmability for shared and distributed memory systems.---Contact:Bruce [email protected] URL: https://fastcode.org/events/fastcode-seminar/mike-spear/---Location: Zoom (register here<https://mit.zoom.us/meeting/register/jsjIZv1HTmSW5Pp02Zxf0g>)  TBD

June 24, 2025

Boltz: Towards a Unified Approach for Biomolecular Interaction Modeling

Gabriele Corso and Jeremy Wohlwend

Part Of

7:00P
- 9:00P

Location

32-G449
Add to Calendar 2025-06-24 19:00:00 2025-06-24 21:00:00 America/New_York Boltz: Towards a Unified Approach for Biomolecular Interaction Modeling 7:00 PM, Tuesday, 24 June 2025MIT Room 32-G449 (Kiva) and online via ZoomBoltz: Towards a Unified Approach for Biomolecular Interaction ModelingGabriele Corso and Jeremy WohlwendPlease register in advance for this seminar even if you plan to attend in person athttps://acm-org.zoom.us/webinar/register/9317463194385/WN_U1yhFblMQO-I4n1cBhkBFwAfter registering, you will receive a confirmation email containing information about joining the webinar.Indicate on the registration form if you plan to attend in person.  This will help us determine whether the room is close to reaching capacity. We plan to serve light refreshments (probably pizza) beforethe talk starting at around 6:30 pm. Letting us know you will come in person will help us determine how much pizza to order.                We may make some auxiliary material such as slides and access to the recording available after the seminar to people who have registered.Abstract:Understanding biomolecular interactions is fundamental to advancing fields like drug discovery and protein design. In this talk, we introduce Boltz-1, an open-source deep learning model incorporatinginnovations in model architecture, speed optimization, and data processing achieving AlphaFold3-level accuracy in predicting the 3D structures of biomolecular complexes. Boltz-1 demonstrates aperformance on-par with state-of-the-art commercial models on a range of diverse benchmarks, setting a new benchmark for commercially accessible tools in structural biology. By releasing the training andinference code, model weights, datasets, and benchmarks under the MIT open license, we aim to foster global collaboration, accelerate discoveries, and provide a robust platform for advancing biomolecularmodeling.Bio: Jeremy Wohlwend and Gabriele Corso are PhD students at the MIT Computer Science and Artificial Intelligence Laboratory where their research focuses on developing novel ML frameworks to tacklechallenging problems in drug discovery and immunology.Directions to 32-G449 - MIT Stata Center, 32 Vassar Street, Cambridge, MA: Please use the main entrance to the Stata Center at 32 Vassar Street (the entrance closest to Main street) as those doors will beunlocked. Upon entering, proceed to the elevators which will be on the right after passing a large set of stairs and a MITAC kiosk. Take the elevator to the 4th floor and turn right, following the hall to anopen area; 32-G449 will be on the left. Location of Stata on campus map.This joint meeting of the Boston Chapter of the IEEE Computer and EMBS Societies and GBC/ACM will be hybrid (in person and online).Up-to-date information about this and other talks is available online at https://ewh.ieee.org/r1/boston/computer/. You can sign up to receive updated status information about this talk and informationalemails about future talks at https://mailman.mit.edu/mailman/listinfo/ieee-cs, our self-administered mailing list.  TBD

September 23, 2025

Explicit Lossless Vertex Expanders

Rachel Zhang
CSAIL, EECS

Part Of

4:15P
- 5:15P

Location

32-G449
Refreshments at 4:00 PM
Add to Calendar 2025-09-23 16:15:00 2025-09-23 17:15:00 America/New_York Explicit Lossless Vertex Expanders We give the first explicit construction of lossless vertex expanders. These are d-regular graphs where every small set S of vertices has (1-eps)d|S| distinct neighbors. Previously, the strongest known explicit vertex expanders were those given by Ramanujan graphs, whose spectral properties imply that every small set S of vertices has 0.5d|S| distinct neighbors.Based on joint work with Jun-Ting Hsieh, Ting-Chun Lin, Alex Lubotzky, Sidhanth Mohanty, Ryan O'Donnell, and Assaf Reiner. TBD