BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Penn Engineering Events - ECPv6.15.18//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-ORIGINAL-URL:https://seasevents.nmsdev7.com
X-WR-CALDESC:Events for Penn Engineering Events
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20240310T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20241103T060000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20250309T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20251102T060000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20260308T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20261101T060000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250306T120000
DTEND;TZID=America/New_York:20250306T131500
DTSTAMP:20260403T135429
CREATED:20250131T195817Z
LAST-MODIFIED:20250131T195817Z
UID:10008257-1741262400-1741266900@seasevents.nmsdev7.com
SUMMARY:IDEAS/STAT Optimization Seminar: "Data-Driven Algorithm Design and Verification for Parametric Convex Optimization"
DESCRIPTION:Zoom link \nhttps://upenn.zoom.us/j/98220304722 \n  \nAbstract\nWe present computational tools for analyzing and designing first-order methods in parametric convex optimization. These methods are popular for their low per-iteration cost and warm-starting capabilities. However\, precisely quantifying the number of iterations required to compute high-quality solutions remains a key challenge\, especially in real-time applications. First\, we introduce a numerical framework for verifying the worst-case performance of first-order methods in parametric quadratic optimization. We formulate this as a mixed-integer linear program that maximizes the infinity norm of the fixed-point residual after a given number of iterations. Our approach captures a broad class of gradient\, projection\, and proximal iterations through affine or piecewise-affine constraints\, with strong polyhedral formulations. To improve scalability\, we incorporate bound-tightening techniques that exploit operator-theoretic bounds. Numerical results show that our method closely matches true worst-case performance\, achieving significant reductions in worst-case fixed-point residuals compared to standard convergence analyses. Second\, we present a data-driven approach for analyzing the performance of first-order methods using statistical learning theory. We establish generalization guarantees for classical optimizers using sample convergence bounds and for learned optimizers using the Probably Approximately Correct (PAC)-Bayes framework. We then apply this framework to learn accelerated first-order methods by directly minimizing the PAC-Bayes bound over key algorithmic parameters (e.g.\, gradient steps and warm-starts). Numerical experiments demonstrate that our approach provides strong generalization guarantees for both classical and learned optimizers\, with statistical bounds that closely match true out-of-sample performance.
URL:https://seasevents.nmsdev7.com/event/ideas-stat-optimization-seminar-bartolomeo-stellato/
LOCATION:Amy Gutmann Hall\, Room 414\, 3333 Chestnut Street\, Philadelphia\, 19104\, United States
CATEGORIES:Seminar,Colloquium
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250306T153000
DTEND;TZID=America/New_York:20250306T163000
DTSTAMP:20260403T135429
CREATED:20250220T194447Z
LAST-MODIFIED:20250220T194447Z
UID:10008300-1741275000-1741278600@seasevents.nmsdev7.com
SUMMARY:CIS Seminar: "Pareto-efficient AI systems: Expanding the quality and efficiency frontier of AI"
DESCRIPTION:We have made exciting progress in AI by massive models on massive amounts of data center compute. However\, the demands for AI are rapidly expanding. I identify how to maximize performance under any compute constraint\, expanding the Pareto frontier of AI capabilities.\n \nThis talk builds up to an efficient language model architecture that expands the Pareto-frontier between quality and throughput efficiency. In motivation\, the Transformer\, AI’s current workhorse architecture\, is memory hungry\, severely limiting its throughput\, or amount of text it can process per second. This has led to a Cambrian explosion of alternate efficient architecture candidates proposed across prior work. Prior work has painted an exciting picture: there exists architectures that are asymptotically faster than Transformers\, while also matching quality. However\, I ask\, if we’re using asymptotically faster building blocks\, are we giving something up in quality?\n\n\nIn part one\, we build understanding. Indeed\, there’s no free lunch! I present my work to identify and explain the fundamental quality and efficiency tradeoffs between different classes of architectures. Methods I developed for this analysis are now ubiquitous in the development of language models.\n\n\nIn part two\, we measure how AI architecture candidates fare on the tradeoff space. A major hurdle\, however\, is that we lack implementations of the architectures that that run at peak-efficiency on modern hardware. Further\, many proposed architectures are asymptotically fast\, but not wall-clock fast. I present ThunderKittens\, a new programming library I built to help AI researchers write simple\, hardware-efficient algorithms across hardware platforms.\n\n\nIn part three\, we expand the Pareto-frontier of the tradeoff space. I present the BASED architecture\, which is built from simple\, hardware-efficient components. I released the state-of-the-art 8B-405B Transformer-free language models\, per standard evaluations\, all on an academic budget.\n\n\nGiven the massive investment into language models\, this work has had significant impact and adoption in research\, open-source\, and industry.
URL:https://seasevents.nmsdev7.com/event/cis-seminar-pareto-efficient-ai-systems-expanding-the-quality-and-efficiency-frontier-of-ai/
LOCATION:Wu and Chen Auditorium (Room 101)\, Levine Hall\, 3330 Walnut Street\, Philadelphia\, PA\, 19104\, United States
ORGANIZER;CN="Computer and Information Science":MAILTO:cherylh@cis.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250306T153000
DTEND;TZID=America/New_York:20250306T163000
DTSTAMP:20260403T135429
CREATED:20250304T142140Z
LAST-MODIFIED:20250304T142140Z
UID:10008312-1741275000-1741278600@seasevents.nmsdev7.com
SUMMARY:Grace Hopper Distinguished Lecture: Melody Swartz
DESCRIPTION:“Immunoregulatory Roles of Lymphatic Vessels in Cancer and Opportunities for Immunoengineering” \nMelody Swartz will discuss her pioneering research on the lymphatic system. This talk will highlight her translational work in immunotherapy\, including lymph node-targeting vaccines and engineering tumor microenvironment models for therapeutic discovery.
URL:https://seasevents.nmsdev7.com/event/grace-hopper-distinguished-lecture-series/
LOCATION:Berger Auditorium (Room 13)\, Skirkanich Hall\, 210 South 33rd Street\, Philadelphia\, PA\, 19104\, United States
CATEGORIES:Distinguished Lecture
ORGANIZER;CN="Bioengineering":MAILTO:be@seas.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250307T100000
DTEND;TZID=America/New_York:20250307T110000
DTSTAMP:20260403T135429
CREATED:20250304T141006Z
LAST-MODIFIED:20250304T141006Z
UID:10008311-1741341600-1741345200@seasevents.nmsdev7.com
SUMMARY:CIS Seminar: "Next Generation Operating Systems for the Cloud"
DESCRIPTION:Modern datacenters must handle an ever-growing array of real-time and data-intensive workloads\, such as interactive web services and AI models\, that demand both low latency and high throughput. However\, traditional operating systems introduce significant I/O overhead\, degrading performance and reducing efficiency. A common solution is to let applications directly communicate with hardware\, bypassing the operating system altogether. While this greatly improves performance\, it sacrifices compatibility with existing software and requires operators to dedicate hardware to each application\, ultimately reducing overall resource utilization. \nIn this talk\, I will present a new datacenter operating system design that achieves high I/O performance without making these tradeoffs. First\, I will introduce Shenango\, which allows applications that directly access hardware to efficiently share CPU cores with other tasks\, maintaining high performance without requiring dedicated resources. Next\, I will discuss Caladan\, a system that prevents performance degradation by managing interference among co-located applications. Finally\, I will describe Junction\, a library operating system that extends these benefits to unmodified applications\, unlocking higher performance and efficiency across diverse workloads.
URL:https://seasevents.nmsdev7.com/event/cis-seminar-next-generation-operating-systems-for-the-cloud/
LOCATION:Amy Gutmann Hall\, Room 414\, 3333 Chestnut Street\, Philadelphia\, 19104\, United States
ORGANIZER;CN="Computer and Information Science":MAILTO:cherylh@cis.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250307T103000
DTEND;TZID=America/New_York:20250307T114500
DTSTAMP:20260403T135429
CREATED:20241213T192945Z
LAST-MODIFIED:20241213T192945Z
UID:10008197-1741343400-1741347900@seasevents.nmsdev7.com
SUMMARY:Spring 2025 GRASP on Robotics: Terry Fong\, NASA Ames Research Center\, “The NASA Volatiles Inspecting Polar Exploration Rover (VIPER) Mission”
DESCRIPTION:This will be a hybrid event with in-person attendance in Wu and Chen and virtual attendance on Zoom. \nABSTRACT\nThe Volatiles Investigating Polar Exploration Rover (VIPER) is a NASA mission designed to explore the extreme environment of the Moon in search of water ice. VIPER is intended to land at the South Pole of the Moon and spend approximately 100-days mapping and surveying four different “ice stability regions”. Determining the distribution\, physical state and composition of water ice deposits will help increase understanding the sources of lunar polar water\, as well as providing insight into the distribution and origin of volatiles across the solar system. \nIn this talk\, I will present an overview of the VIPER mission\, the rover’s design\, lunar surface simulation\, and mission operations. During VIPER’s exploration of the Moon\, the rover will endure extreme temperature conditions\, dynamic lighting and complex terrain\, while near-real-time rover driving will present new planetary mission operational challenges.
URL:https://seasevents.nmsdev7.com/event/spring-2025-grasp-on-robotics-terry-fong-nasa-ames-research-center-the-nasa-volatiles-inspecting-polar-exploration-rover-viper-mission/
LOCATION:Wu and Chen Auditorium (Room 101)\, Levine Hall\, 3330 Walnut Street\, Philadelphia\, PA\, 19104\, United States
ORGANIZER;CN="General Robotics%2C Automation%2C Sensing and Perception (GRASP) Lab":MAILTO:grasplab@seas.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250311T101500
DTEND;TZID=America/New_York:20250311T111500
DTSTAMP:20260403T135429
CREATED:20250225T222029Z
LAST-MODIFIED:20250225T222029Z
UID:10008305-1741688100-1741691700@seasevents.nmsdev7.com
SUMMARY:MEAM Seminar: "Three-Dimensional Biointerfaces: Soft Bioelectronics for Complex Biological Geometries"
DESCRIPTION:Cutting-edge biological and medical research demands innovative methods for sensing and modulating complex tissues\, organs\, and organ systems. Recent progress in bioelectronics enables multimodal interfacing for broad fundamental and therapeutic applications. However\, key challenges persist in interfacing with complex biological geometries\, particularly for applications requiring conformal contact in electrical\, chemical\, or mechanical biointerfaces. In this talk\, I will introduce our work on three-dimensional biointerfaces to address this challenge. These 3D biointerfaces feature precisely programmable geometries and/or ultrasoft nature that enable conformal contacting with multiscale tissues and organs. The first part of the talk focuses on our work in full-surface 3D interface for 3D-shaped neural organoids. These 3D programmable frameworks enable nearly complete surface accessibility of neuronal populations for the first time. They support high-resolution recording and stimulation via hundreds of individual channels and allow for 3D reconstruction for spatial electrophysiology. These capabilities enable monitoring 3D network-level neural activities\, supporting broad biological and medical research as demonstrated with neural activity manipulation\, pharmacological studies\, and modeling neural disease phenotypes. Next\, I will discuss our work on organ interface. We develop hydrogel-based ultrasoft thin platforms that enable real-time pH sensing. These platforms integrate ultrasound imaging for deep tissue monitoring\, and address the long-standing clinical challenge of early detection of gastrointestinal (GI) system leakage after surgery. In the last part of the talk\, I will present our work on novel nanofabrication techniques for large scale and aligned 3D nanostructured surfaces which match the scale of cellular and subcellular sensing. Together\, these works open new possibilities for precision engineering of advanced bioelectronics with 3D features that allow sufficient interfacing with complex geometries\, enabling transformative applications in organoid research and neural engineering\, precision medicine\, and spatial characterization platforms for complex in vitro and in vivo biological systems for advanced medical and therapeutic applications.
URL:https://seasevents.nmsdev7.com/event/meam-seminar-three-dimensional-biointerfaces-soft-bioelectronics-for-complex-biological-geometries/
LOCATION:Wu and Chen Auditorium (Room 101)\, Levine Hall\, 3330 Walnut Street\, Philadelphia\, PA\, 19104\, United States
CATEGORIES:Seminar
ORGANIZER;CN="Mechanical Engineering and Applied Mechanics":MAILTO:meam@seas.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250312T103000
DTEND;TZID=America/New_York:20250312T120000
DTSTAMP:20260403T135429
CREATED:20250212T181014Z
LAST-MODIFIED:20250212T181014Z
UID:10008286-1741775400-1741780800@seasevents.nmsdev7.com
SUMMARY:Seminar: "Merging Bio-integrated Materials and Devices with Ultrasound: Opportunities in Sensing\, Modulation and Actuation" - Dr. Jiaqi Liu\, Ph.D.
DESCRIPTION:Soft\, bio-integrated materials and devices exhibit the ability to collect physiological signals and offer therapeutic functions\, posing a substantial impact on our understanding of biology and paving the way for precision medicine. With the assistance of advanced ultrasound technology\, the innovative integration holds immense promise for sensing\, modulation and actuation in deep tissues and complex anatomical structures that are otherwise difficult to approach. In this talk\, I will highlight our recent advances in bioresorbable\, shape-adaptive materials structures that enable real-time monitoring of deep-tissue homeostasis using ultrasound instruments. I will showcase another example of a bioresorbable implant based on radio-frequency coupling for chemical sensing. I will then summarize our previous efforts in materials design that lead to programmable dual responsive soft actuators. Finally\, I will discuss opportunities in leveraging functional soft materials\, ultrasound imaging and its various forms of energy for measuring biosignals\, provide targeted modulation and programmable actuation to decipher biological systems and improve healthcare.
URL:https://seasevents.nmsdev7.com/event/seminar-merging-bio-integrated-materials-and-devices-with-ultrasound-opportunities-in-sensing-modulation-and-actuation-dr-jiaqi-liu-ph-d/
LOCATION:LRSM Reading Room\, 3231 Walnut St.\, Philadelphia\, PA\, 19104\, United States
CATEGORIES:Seminar
ORGANIZER;CN="Materials Science and Engineering":MAILTO:johnruss@seas.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250313T143000
DTEND;TZID=America/New_York:20250313T153000
DTSTAMP:20260403T135429
CREATED:20250305T185928Z
LAST-MODIFIED:20250305T185928Z
UID:10008313-1741876200-1741879800@seasevents.nmsdev7.com
SUMMARY:Spring 2025 GRASP Seminar: Shubham Tulsiani\, Carnegie Mellon University\, "Reconstructing and Generating 3D"
DESCRIPTION:This will be a hybrid event with in-person attendance in Raisler Lounge (Towne 225) and virtual attendance on Zoom. \nABSTRACT\nModern reconstruction systems\, leveraging advances in pose estimation and generative modeling\, can reconstruct everyday objects and scenes from casually captured images. In this talk\, I will highlight recent work from our group that pushes these frontiers. I will first present a learning-based approach to the structure-from-motion task\, highlighting how we can rethink camera parametrization for neural prediction methods. Building on this\, I will describe a framework for combining generative priors and pose estimation for high-fidelity 3D reconstruction ‘in-the-wild’. Finally\, I will outline an approach for ‘ultra-fast’ 3D generation that allows users to synthesize high-fidelity 3D assets in under a second.
URL:https://seasevents.nmsdev7.com/event/spring-2025-grasp-seminar-shubham-tulsiani-carnegie-mellon-university-reconstructing-and-generating-3d/
LOCATION:Raisler Lounge (Room 225)\, Towne Building\, 220 South 33rd Street\, Philadelphia\, PA\, 19104\, United States
CATEGORIES:Seminar
ORGANIZER;CN="General Robotics%2C Automation%2C Sensing and Perception (GRASP) Lab":MAILTO:grasplab@seas.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250313T153000
DTEND;TZID=America/New_York:20250313T163000
DTSTAMP:20260403T135429
CREATED:20250205T192804Z
LAST-MODIFIED:20250205T192804Z
UID:10008270-1741879800-1741883400@seasevents.nmsdev7.com
SUMMARY:BE Seminar - “Probing nuclear structure and function with novel genomic tools” (Sofia Quinodoz\, Princeton University)
DESCRIPTION:
URL:https://seasevents.nmsdev7.com/event/be-seminar-probing-nuclear-structure-and-function-with-novel-genomic-tools-sofia-quinadoz-princeton-university/
LOCATION:Berger Auditorium (Room 13)\, Skirkanich Hall\, 210 South 33rd Street\, Philadelphia\, PA\, 19104\, United States
CATEGORIES:Seminar
ATTACH;FMTTYPE=image/png:https://seasevents.nmsdev7.com/wp-content/uploads/2025/02/Quinodoz_Sofia_Headshot17159514.png
ORGANIZER;CN="Bioengineering":MAILTO:be@seas.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250318T101500
DTEND;TZID=America/New_York:20250318T111500
DTSTAMP:20260403T135429
CREATED:20241231T195411Z
LAST-MODIFIED:20241231T195411Z
UID:10008206-1742292900-1742296500@seasevents.nmsdev7.com
SUMMARY:MEAM Seminar: "Bioinspired Robotic Sensorimotor Systems via 3D Printed Soft and Architected Materials"
DESCRIPTION:Animals can easily adapt their bodies and movements to new\, unstructured environments and situations. Robots cannot. While engineers construct robots from rigid\, motorized mechanisms to precisely control their movements\, vertebrates leverage compliant\, deformable musculoskeletal systems to adaptively navigate complex environments and produce dynamically stable gaits and motions. Providing robots with an equivalent musculoskeletal system will open new opportunities for achieving bioinspired motility\, adaptability\, robustness\, and performance. However\, this vision is stymied not only by limitations in current materials and manufacturing methods\, but also in how to strategically integrate soft and rigid materials in robot bodies. With these challenges in mind\, I will present approaches for engineering artificial musculoskeletal systems via 3D printed soft and architected materials. First\, I will introduce a flexible\, architected soft actuator unit for motorized extensional motion. I will introduce techniques for sensorizing these architected actuators\, assembling them for locomoting soft robots\, and interfacing them with skeletal elements for force transmission. Finally\, I will discuss new strategies for architecting soft ionic conductors for distributed sensing. These efforts aim to embody both physical and computational intelligence into real-world-deployable soft robots with practical task capabilities.
URL:https://seasevents.nmsdev7.com/event/meam-seminar-bioinspired-robotic-sensorimotor-systems-via-3d-printed-soft-and-architected-materials/
LOCATION:Wu and Chen Auditorium (Room 101)\, Levine Hall\, 3330 Walnut Street\, Philadelphia\, PA\, 19104\, United States
CATEGORIES:Seminar
ORGANIZER;CN="Mechanical Engineering and Applied Mechanics":MAILTO:meam@seas.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250318T110000
DTEND;TZID=America/New_York:20250318T120000
DTSTAMP:20260403T135429
CREATED:20250206T140537Z
LAST-MODIFIED:20250206T140537Z
UID:10008273-1742295600-1742299200@seasevents.nmsdev7.com
SUMMARY:ESE Spring Seminar - "Generalization\, Memorization\, and Privacy in Trustworthy Machine Learning"
DESCRIPTION:Machine learning is transforming numerous aspects of modern society\, and its expanding use in high-stakes applications calls for responsible development. In this talk\, I will present my research on the foundations and methodologies for building trustworthy ML\, centered on three interconnected challenges: generalization\, memorization\, and privacy. First\, I will show how information-theoretic tools can be used to analyze generalization across different learning setups. Next\, I will describe my work on the fundamental limits of memorization in certain high-dimensional convex settings\, showing a precise trade-off between memorization and accuracy. Finally\, I will propose adaptive and efficient optimization algorithms under differential privacy—a well-established framework designed to protect sensitive data and limit memorization risk—that adapt to the properties of the dataset\, resulting in smaller error. My results highlight how these three pillars interact\, and I will conclude by outlining my plans for future research.
URL:https://seasevents.nmsdev7.com/event/ese-spring-seminar-title-tba/
LOCATION:Raisler Lounge (Room 225)\, Towne Building\, 220 South 33rd Street\, Philadelphia\, PA\, 19104\, United States
CATEGORIES:Seminar,Colloquium
ORGANIZER;CN="Electrical and Systems Engineering":MAILTO:eseevents@seas.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250318T153000
DTEND;TZID=America/New_York:20250318T163000
DTSTAMP:20260403T135429
CREATED:20250313T165613Z
LAST-MODIFIED:20250313T165613Z
UID:10008317-1742311800-1742315400@seasevents.nmsdev7.com
SUMMARY:CIS Seminar: "Probabilistic Experimental Design for Petascale DNA Synthesis"
DESCRIPTION:Generative modeling offers a powerful paradigm for designing novel functional DNA\, RNA and protein sequences. In this talk\, I introduce probabilistic experimental design methods to efficiently manufacture samples from generative models of biomolecules in the real world. These algorithms merge computational techniques for approximate sampling with physical randomness. I also develop tools to rigorously evaluate the quality of manufactured samples\, including nonparametric\ntwo-sample tests with consistency guarantees and scalable algorithms. I demonstrate synthesizing ~10^16 samples from a generative model of human antibodies\, at a sample quality comparable to state-of-the-art protein language models\, and a cost of ~$10^3. The library yields candidate therapeutics for “undruggable” cancer targets. Using previous methods\, manufacturing a DNA library of the same size and quality would cost roughly ~$10^15.
URL:https://seasevents.nmsdev7.com/event/13596/
LOCATION:Amy Gutmann Hall\, Room 414\, 3333 Chestnut Street\, Philadelphia\, 19104\, United States
ORGANIZER;CN="Computer and Information Science":MAILTO:cherylh@cis.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250319T120000
DTEND;TZID=America/New_York:20250319T131500
DTSTAMP:20260403T135429
CREATED:20250205T162229Z
LAST-MODIFIED:20250205T162229Z
UID:10008269-1742385600-1742390100@seasevents.nmsdev7.com
SUMMARY:ASSET Seminar: "Algorithmic Stability for Trustworthy Machine Learning and Statistics"
DESCRIPTION:Abstract: \nData-driven systems hold immense potential to positively impact society\, but their reliability remains a challenge. Their outputs are often too brittle to changes in their training data\, leaving them vulnerable to data poisoning attacks\, prone to leaking sensitive information\, or susceptible to overfitting. Establishing fundamental principles for designing algorithms that are both stable—to mitigate these risks—and efficient in their use of resources is essential for enabling trustworthy data-driven systems. \nIn this talk\, I will focus on statistical estimation under differential privacy—a rigorous framework that ensures data-driven system outputs do not reveal sensitive information about individuals in their input. I will present algorithmic techniques that take advantage of beneficial structure in the data to achieve optimal error for several multivariate tasks without requiring any prior information about the data\, by building on robustness against data poisoning attacks. Lastly\, I will highlight the deeper connection between differential privacy and robustness that underpins these results. \nZoom Link (if unable to attend in-person): https://upenn.zoom.us/j/94645455801
URL:https://seasevents.nmsdev7.com/event/asset-seminar-lydia-zakynthinou-uc-berkeley/
LOCATION:Amy Gutmann Hall\, Room 414\, 3333 Chestnut Street\, Philadelphia\, 19104\, United States
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250319T150000
DTEND;TZID=America/New_York:20250319T160000
DTSTAMP:20260403T135429
CREATED:20250312T153849Z
LAST-MODIFIED:20250312T153849Z
UID:10008316-1742396400-1742400000@seasevents.nmsdev7.com
SUMMARY:Spring 2025 GRASP SFI: Student Lightning Talks\, Session 2
DESCRIPTION:This will be a hybrid event with in-person attendance in Levine 307 and virtual attendance on Zoom. \nABSTRACT\nThe Spring 2025 GRASP SFI Student Lightning Talks will highlight the research of three GRASP Lab Master’s or early PhD students whose presentation topics have been nominated by their faculty advisors and voted on by their GRASP peers. \nAhmad Amine (PhD\, ESE) \nSIT-LMPC: Safe Information-Theoretic Learning Model Predictive Control for Iterative Tasks \nIterative tasks are common in robotics applications ranging from pick-and-place tasks to autonomous racing. Finding optimal trajectories for these tasks has been thoroughly explored in literature through techniques like trajectory planning\, optimal control\, and reinforcement learning. Learning Model Predictive Control (LMPC) solves for the optimal trajectory that maximizes performance for a given task by learning the value function from trajectories executed in previous iterations. Unlike reinforcement learning\, LMPC provides desirable theoretical properties for safety and convergence\, which is crucial for safety-critical applications. We present an approach for extending the LMPC problem to stochastic nonlinear systems using Model Predictive Path Integral (MPPI). Previous methods utilizing optimization-based techniques or Cross-Entropy-Method sampling to solve the LMPC problem fall short when dealing with high-dimensional\, nonlinear dynamics. Our MPPI-based framework optimizes the control policy from an information-theoretic perspective and overcomes these limitations by providing a systematic way of handling constraints without sacrificing sample spread. This is achieved by incorporating penalty functions and optimizing over multipliers to balance out constraint satisfaction with controller performance. The proposed algorithm can be parallelized allowing for real-time control on limited compute hardware. We validate our approach through simulations and real-world experiments\, demonstrating significant improvements in constraint satisfaction and final trajectory performance. \nErica Santos (MSE\, ROBO) \nTemporarily\, Robots Unite to Surmount Sandy Entrapments\, then Separate (TRUSSES) \nThe TRUSSES project is developing methods for teams of robots to jointly overcome environmental hazards on the Moon by attaching to each other to form larger and more stable\, maneuverable structures. The robots will use their interactions with the ground to form a map of safe and risky terrain\, attach to each other as support when the ground traversal risk is high\, move in a coordinated fashion once joined\, and\, once the maneuver has been successfully completed\, separate to continue their original individual missions. \nDarshan Thaker (PhD\, CIS) \nFrequency Guided Posterior Sampling for Diffusion-based Image Restoration \nImage restoration aims to recover high-quality images from degraded observations. When the degradation process is known\, the recovery problem can be formulated as an inverse problem\, and in a Bayesian context\, the goal is to sample a clean reconstruction given the degraded observation. Recently\, modern pretrained diffusion models have been used for image restoration by modifying their sampling procedure to account for the degradation process. However\, these methods often rely on certain approximations that can lead to significant errors and compromised sample quality. We present a simple modification to existing diffusion-based restoration methods that helps mitigate these errors based on a frequency analysis of the degraded observations. This results in significant improvement on challenging image restoration tasks such as motion deblurring and image dehazing.
URL:https://seasevents.nmsdev7.com/event/spring-2025-grasp-sfi-student-lightning-talks-session-2/
LOCATION:Levine 307\, 3330 Walnut Street\, Philadelphia\, PA\, 19104\, United States
CATEGORIES:Seminar
ORGANIZER;CN="General Robotics%2C Automation%2C Sensing and Perception (GRASP) Lab":MAILTO:grasplab@seas.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250319T153000
DTEND;TZID=America/New_York:20250319T163000
DTSTAMP:20260403T135429
CREATED:20250108T144502Z
LAST-MODIFIED:20250108T144502Z
UID:10008208-1742398200-1742401800@seasevents.nmsdev7.com
SUMMARY:The Harold Berger Distinguished Award Lecture 2025
DESCRIPTION:Quantum Dots: From Curiosity to Technological Impact\nThe combination of quantum effects\, nanometer dimensions and a chemical synthesis make quantum dots a platform for exploring new size-dependent fundamental properties and a sandbox for developing new applications. This talk will cover the origin story of chemically synthesized quantum dots\, their basic physics\, the synthesis that unlocked their widespread study and applications\, an overview of their properties and a few of their applications\, including in bio-imaging and as emissive materials in displays. \nRegister Now
URL:https://seasevents.nmsdev7.com/event/the-harold-berger-distinguished-award-lecture-2025/
LOCATION:Wu & Chen Auditorium
CATEGORIES:Distinguished Lecture
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250320T103000
DTEND;TZID=America/New_York:20250320T120000
DTSTAMP:20260403T135429
CREATED:20250221T195925Z
LAST-MODIFIED:20250221T195925Z
UID:10008302-1742466600-1742472000@seasevents.nmsdev7.com
SUMMARY:MSE Seminar: "Soft Microparticle Assemblies to MAP Scaffolds" Tatiana Segura - Duke University
DESCRIPTION:Particle assemblies form interconnected pockets of empty space that are hot spots for activity in many applications and natural phenomenon that deal with particulate matter. In my lab we explore the use of interlinked soft microparticle assemblies as scaffolds for cell culture in vitro and to promote regenerative wound healing in vivo. The open space that is created by these soft microparticle assemblies can be engineered to guide cellular behavior through restricting the space occupied by cells by modulating particle size\, changing the chemical composition of the open space\, and activating the cells with soluble factors. Using these approaches\, we have studied the role of open space on macrophage polarization\, neuroprogenitor cell growth\, and vascular morphogenesis. Injection of these materials into wounds result in the local modulation of immune cells that turn the environment towards a reparative environment and leads to scarless healing skin. In this talk\, I will discuss our latest results in our creation of soft particle assemblies for biomedical applications.
URL:https://seasevents.nmsdev7.com/event/mse-seminar-soft-microparticle-assembles-to-map-scaffolds-tatiana-segura-duke-university/
LOCATION:Wu & Chen Auditorium
CATEGORIES:Seminar
ORGANIZER;CN="Materials Science and Engineering":MAILTO:johnruss@seas.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250320T110000
DTEND;TZID=America/New_York:20250320T120000
DTSTAMP:20260403T135429
CREATED:20250303T184346Z
LAST-MODIFIED:20250303T184346Z
UID:10008310-1742468400-1742472000@seasevents.nmsdev7.com
SUMMARY:ESE Spring Seminar - "Can Robots Learn from Machine Dreams? - Robot Learning via GenAI-powered World Models"
DESCRIPTION:Over the past decade\, large-scale pre-training followed by alignment has revolutionized natural language processing and computer vision. Yet\, robotics remains constrained by the scarcity of real-world data. In this talk\, I will present our systematic approach to overcoming this bottleneck by building increasingly rich world models from data. I will first introduce our distilled feature field work that represents a robot’s action in relation to multi-modal descriptors in a static environment. Building upon this foundation\, I will demonstrate LucidSim\, a GenAI-powered simulator that significantly narrows the sim-to-real gap in the visual domain. I will show the initial steps we have taken toward a robot foundation model by extending this framework from legged locomotion to manipulation tasks. These results suggest that synthetic data sourced from world models can accelerate robot learning beyond what is possible with real-world data alone. I will conclude by outlining a roadmap that begins with enabling machines to dream a million worlds that will ultimately allow AI to enter the physical world.
URL:https://seasevents.nmsdev7.com/event/ese-spring-seminar-title-tba-5/
LOCATION:Raisler Lounge (Room 225)\, Towne Building\, 220 South 33rd Street\, Philadelphia\, PA\, 19104\, United States
CATEGORIES:Seminar,Colloquium
ORGANIZER;CN="Electrical and Systems Engineering":MAILTO:eseevents@seas.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250320T120000
DTEND;TZID=America/New_York:20250320T131500
DTSTAMP:20260403T135429
CREATED:20250131T195937Z
LAST-MODIFIED:20250131T195937Z
UID:10008258-1742472000-1742476500@seasevents.nmsdev7.com
SUMMARY:IDEAS/STAT Optimization Seminar: "Statistics-Powered ML: Building Trust and Robustness in Black-Box Predictions"
DESCRIPTION:Zoom link: \nhttps://upenn.zoom.us/j/98220304722 \nAbstract: \nModern ML models produce valuable predictions across various applications\, influencing people’s lives\, opportunities\, and scientific advancements. However\, these systems can fail in unexpected ways\, generating unreliable inferences and perpetuating biases present in the data. These issues are particularly troubling in high-stakes applications\, where models are trained on increasingly diverse\, incomplete\, and noisy data and then deployed in dynamic environments—conditions that often exacerbate test-time failures. \nIn response to these challenges\, this talk explores a key question: How can fundamental statistical principles be harnessed to produce trustworthy predictive inference? \nIn the first part\, I will present a new advancement in conformal prediction—a statistical wrapper for any black-box model that provides precise error bounds on ML predictions. I will focus on scenarios where training data is corrupted or biased\, such as through missing features and labels\, and introduce a framework for constructing predictive uncertainty estimates that remain valid despite distribution shifts between the available corrupted data and unknown clean data. \nIn the second part\, I will show how sequential statistical testing can enable a novel test-time training scheme\, allowing a pre-trained model to adapt online to unfamiliar environments. For instance\, consider an image classification task where test images are captured under varying illumination conditions that differ from the training setup. Building on conformal betting martingales\, I will first introduce a monitoring tool to detect data drifts. Using this tool\, I will derive a rigorous ‘anti-drift correction’ mechanism grounded in (online) optimal transport principles. This mechanism forms the foundation of a self-training scheme that produces robust predictions invariant to dynamically changing environments.
URL:https://seasevents.nmsdev7.com/event/ideas-stat-optimization-seminar-pratik-chaudhari/
LOCATION:Amy Gutmann Hall\, Room 414\, 3333 Chestnut Street\, Philadelphia\, 19104\, United States
CATEGORIES:Seminar,Colloquium
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250320T153000
DTEND;TZID=America/New_York:20250320T163000
DTSTAMP:20260403T135429
CREATED:20250313T171704Z
LAST-MODIFIED:20250313T171704Z
UID:10008318-1742484600-1742488200@seasevents.nmsdev7.com
SUMMARY:CIS Seminar: "AI for Materials Discovery: Graphs\, Language Models\, and Agents"
DESCRIPTION:Artificial intelligence (AI) is transforming scientific discovery\, particularly in materials science\, by accelerating the prediction and design of materials with desired properties. Traditional physics-based modeling of atomic systems is computationally prohibitive for large-scale problems\, and AI addresses this challenge by learning the underlying physics from data\, thereby accelerating discoveries. In this talk I will present advances in AI-driven materials discovery\, focusing on integrating physical principles\, such as symmetry and equivariance\, into AI models for accurate prediction and generation. My key approaches include geometric deep learning\, language models\, and AI agents\, which collectively enhance the efficiency of materials discovery and strengthen the connection between AI and physical sciences. I will discuss my ongoing and future work\, aiming at developing foundation models for materials and generic atomic systems\, along with automated AI-driven discovery pipelines. I will conclude by presenting my future plans in tackling more complex and multiscale systems\, spanning disordered materials and high-entropy alloys to general AI for science problems. Altogether\, my current and future research demonstrates the close synergy among AI\, physics\, chemistry\, and materials science.
URL:https://seasevents.nmsdev7.com/event/cis-seminar-ai-for-materials-discovery-graphs-language-models-and-agents/
LOCATION:Levine 307\, 3330 Walnut Street\, Philadelphia\, PA\, 19104\, United States
ORGANIZER;CN="Computer and Information Science":MAILTO:cherylh@cis.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250325T101500
DTEND;TZID=America/New_York:20250325T111500
DTSTAMP:20260403T135429
CREATED:20241126T215236Z
LAST-MODIFIED:20241126T215236Z
UID:10008188-1742897700-1742901300@seasevents.nmsdev7.com
SUMMARY:MEAM Seminar: "Multiscale and Multi-physics Mechanics Involving Highly Flexible Nano and Continuum Rods"
DESCRIPTION:Flexible rod-like structures such as nanotubes and nanowires have found significant interest in nanoelectronic applications. Likewise\, with the recent advancement in soft robotics and additive manufacturing\, an important goal is to optimally design architected slender metamaterials and further derive their effective mechanical properties through homogenization techniques. However\, modeling such structures as a flexible continuum body poses multiscale challenges. Lately\, there is also a surge in interest for understanding the motion and deformation of such slender structures in the presence of surrounding fluid or external electric/magnetic field. In this talk\, I will touch upon the mechanics and numerical challenges faced while addressing the above mentioned problems.
URL:https://seasevents.nmsdev7.com/event/meam-seminar-multiscale-and-multi-physics-mechanics-involving-highly-flexible-nano-and-continuum-rods/
LOCATION:Wu and Chen Auditorium (Room 101)\, Levine Hall\, 3330 Walnut Street\, Philadelphia\, PA\, 19104\, United States
CATEGORIES:Seminar
ORGANIZER;CN="Mechanical Engineering and Applied Mechanics":MAILTO:meam@seas.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250325T110000
DTEND;TZID=America/New_York:20250325T120000
DTSTAMP:20260403T135429
CREATED:20250225T175645Z
LAST-MODIFIED:20250225T175645Z
UID:10008304-1742900400-1742904000@seasevents.nmsdev7.com
SUMMARY:ESE Spring Seminar - "Engineering and utilizing interactions between spins and light with molecular qubits"
DESCRIPTION:Coupling of spins and light can enable photon-mediated scaling and control in quantum technologies\, as demonstrated in trapped atom\, ion\, and solid-state spin qubits. Molecular analogs of such systems hold promise as a nascent qubit platform that can leverage the tools of synthetic chemistry to tailor quantum properties and integrate in nanoscale devices but have thus far been limited by broad optical transitions. In this talk I describe the development of optically addressable molecular qubits utilizing a central spin-bearing rare-earth ion. We demonstrate coherent microwave control of the spin ground-state and high-resolution spin-selective optical transitions. The resulting spin-optical interface enables high-contrast detection and optical generation of non-equilibrium spin polarization. This approach opens new doors for engineering and utilizing spin-photon coupling in heterogeneous molecular devices with capacity for chemical design.
URL:https://seasevents.nmsdev7.com/event/ese-spring-seminar-title-tba-4/
LOCATION:Raisler Lounge (Room 225)\, Towne Building\, 220 South 33rd Street\, Philadelphia\, PA\, 19104\, United States
CATEGORIES:Seminar,Colloquium
ORGANIZER;CN="Electrical and Systems Engineering":MAILTO:eseevents@seas.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250325T153000
DTEND;TZID=America/New_York:20250325T163000
DTSTAMP:20260403T135429
CREATED:20250319T135310Z
LAST-MODIFIED:20250319T135310Z
UID:10008320-1742916600-1742920200@seasevents.nmsdev7.com
SUMMARY:CIS Seminar presents: " Cyber-Physical Security Through the Lens of AI-Enabled Systems
DESCRIPTION:Cyber-physical systems (CPS)\, powered by emerging artificial intelligence (AI) technologies\, have become integral to various critical domains such as the Internet of Things (IoTs)\, medical devices\, and autonomous vehicles. A unique aspect of these systems lies in their interactions with the physical world\, by perceiving environments through heterogeneous modalities (perception)\, processing digital data with intelligence algorithms (computing)\, and autonomously actuating controls that affect physical processes (actuation). While this intricate fusion of cyber and physical components has unlocked unprecedented capabilities\, it has also introduced new security challenges. However\, traditional security measures often fall short in addressing these multifaceted threats. Under this paradigm shift\, I systematically explore and mitigate the threats inherent in AI-enabled cyber-physical systems. The research objectives are threefold: (1) investigating how the interplay of cyber and physical components opens up novel attack and defense vectors\, (2) developing robust defense strategies grounded by physical laws and reasoning\, and (3) benchmarking and theoretically analyzing security trade-offs from algorithmic\, system-level\, and human-centric perspectives. By bridging the gap between cyber and physical domains\, my research enhances the resilience and trustworthiness of modern CPS while retaining system efficiency and usability.
URL:https://seasevents.nmsdev7.com/event/cis-seminar-presents-cyber-physical-security-through-the-lens-of-ai-enabled-systems/
LOCATION:Amy Gutmann Hall\, Room 414\, 3333 Chestnut Street\, Philadelphia\, 19104\, United States
ORGANIZER;CN="Computer and Information Science":MAILTO:cherylh@cis.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250326T120000
DTEND;TZID=America/New_York:20250326T131500
DTSTAMP:20260403T135429
CREATED:20250128T132649Z
LAST-MODIFIED:20250128T132649Z
UID:10008241-1742990400-1742994900@seasevents.nmsdev7.com
SUMMARY:ASSET Seminar: "Controlling Language Models"
DESCRIPTION:Abstract: \nControlling language models is key to unlocking their full potential and making them useful for downstream tasks. Successfully deploying these models often requires both task-specific customization and rigorous auditing of their behavior. In this talk\, I will begin by introducing a customization method called Prefix-Tuning\, which adapts language models by updating only 0.1% of their parameters. Next\, I will address the need for robust auditing by presenting a Frank-Wolfe-inspired algorithm for red-teaming language models\, which provides a principled framework for discovering diverse failure modes. Finally\, I will rethink the root cause of these control challenges\, and propose a new generative model for text\, called Diffusion-LM\, which is controllable by design. \nZoom Link (if unable to attend in-person): https://upenn.zoom.us/j/93867005722
URL:https://seasevents.nmsdev7.com/event/asset-seminar-lisa-li-stanford-university/
LOCATION:Amy Gutmann Hall\, Room 414\, 3333 Chestnut Street\, Philadelphia\, 19104\, United States
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250326T150000
DTEND;TZID=America/New_York:20250326T160000
DTSTAMP:20260403T135429
CREATED:20250306T164843Z
LAST-MODIFIED:20250306T164843Z
UID:10008315-1743001200-1743004800@seasevents.nmsdev7.com
SUMMARY:Spring 2025 GRASP SFI: Danna Ma\, Cornell University\, “Harnessing Physical Intelligence for Collective Motion in Robotic Matter”
DESCRIPTION:This will be a hybrid event with in-person attendance in Levine 307 and virtual attendance on Zoom. \nABSTRACT\nIn recent years\, the field of swarm robotics has seen rapid advancements\, from research to industry. While most work focuses on programmed intelligence\, swarms in nature demonstrate that physical intelligence—where agents perform tasks based on their morphology and physical interactions with the environment—also plays a critical role. In my work\, I have developed a robotic swarm of smart particles\, or ‘smarticles\,’ which can serve both as a taskable robotic platform and as a smart material. In this talk\, I will discuss the electro-mechanical design\, sensory feedback mechanisms\, and adaptive control strategies that enable smarticles to exhibit emergent collective motion\, demonstrating how they can operate autonomously without the need for a central controller\, high-bandwidth communication\, or extensive pre-planning.
URL:https://seasevents.nmsdev7.com/event/spring-2025-grasp-sfi-danna-ma/
LOCATION:Levine 307\, 3330 Walnut Street\, Philadelphia\, PA\, 19104\, United States
CATEGORIES:Seminar
ORGANIZER;CN="General Robotics%2C Automation%2C Sensing and Perception (GRASP) Lab":MAILTO:grasplab@seas.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250327T110000
DTEND;TZID=America/New_York:20250327T120000
DTSTAMP:20260403T135429
CREATED:20250213T183414Z
LAST-MODIFIED:20250213T183414Z
UID:10008287-1743073200-1743076800@seasevents.nmsdev7.com
SUMMARY:ESE Spring Seminar - "Wavelength-Encoded Nanolaser Particles for Highly Multiplexed Single-Cell Analysis"
DESCRIPTION:Understanding single-cell heterogeneity in biological systems is considered the holy grail of biomedicine. However\, conventional single-cell analysis methods are constrained by the destructive readout process of DNA barcodes and the broad emission linewidths of fluorescence barcodes\, limiting their ability to capture dynamic information and achieve high multiplexing capabilities. \nThis seminar explores the transformative potential of “nanolaser particles.” By leveraging the broad gain bandwidths of III-V semiconductor materials\, these nanolaser particles generate ultra-narrowband stimulated emissions (<1 nm)\, enabling high-throughput combinatorial living cell barcoding on a scale of millions. \nKey advancements include: (i) large-scale fabrication of injectable\, biocompatible nanolaser particles; (ii) the invention of Laser Particle Stimulated Emission Microscopy (LASE) for super-resolution\, real-time in vitro and in vivo cell tracking\, as well as spatial transcriptomics of migrating cells; and (iii) the development of the world’s smallest laser (170 nm)\, achieving the fundamental half-wave size limit and paving the way for molecular-sized laser probes.
URL:https://seasevents.nmsdev7.com/event/ese-spring-seminar-title-tba-3/
LOCATION:Raisler Lounge (Room 225)\, Towne Building\, 220 South 33rd Street\, Philadelphia\, PA\, 19104\, United States
CATEGORIES:Seminar,Colloquium
ORGANIZER;CN="Electrical and Systems Engineering":MAILTO:eseevents@seas.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250327T120000
DTEND;TZID=America/New_York:20250327T131500
DTSTAMP:20260403T135429
CREATED:20250131T200843Z
LAST-MODIFIED:20250131T200843Z
UID:10008259-1743076800-1743081300@seasevents.nmsdev7.com
SUMMARY:IDEAS/STAT Optimization Seminar: "The Size of Teachers as a Measure of Data Complexity: PAC-Bayes Excess Risk Bounds and Scaling Laws"
DESCRIPTION:Zoom link: https://upenn.zoom.us/j/98220304722 \nAbstract:\nWe study the generalization properties of neural networks through the lens of data complexity.  Recent work by Buzaglo et al. (2024) shows that random (nearly) interpolating networks generalize\, provided there is a small “teacher” network that achieves small excess risk. We give a short single-sample PAC-Bayes proof of this result and an analogous “fast-rate” result for random samples from Gibbs posteriors. The resulting oracle inequality motivates a new notion of data complexity\, based on the minimal size of a teacher network required to achieve any given level of excess risk. We show that polynomial data complexity gives rise to power laws connecting risk to the number of training samples\, like in empirical neural scaling laws. By comparing the “scaling laws” resulting from our bounds to those observed in empirical studies\, we provide evidence for lower bounds on the data complexity of standard benchmarks.\n\nJoint work with G. K. Dziugaite.
URL:https://seasevents.nmsdev7.com/event/ideas-stat-optimization-seminar-dan-roy/
LOCATION:Amy Gutmann Hall\, Room 414\, 3333 Chestnut Street\, Philadelphia\, 19104\, United States
CATEGORIES:Seminar,Colloquium
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250327T153000
DTEND;TZID=America/New_York:20250327T163000
DTSTAMP:20260403T135429
CREATED:20250319T173957Z
LAST-MODIFIED:20250319T173957Z
UID:10008321-1743089400-1743093000@seasevents.nmsdev7.com
SUMMARY:CIS Seminar: "Unlocking Scalable Robot Learning in the Real World"
DESCRIPTION:Many domains of machine learning\, from language modeling to computer vision\, have recently undergone a shift towards generalist models\, whose broad generalization abilities are fueled by large and diverse real-world training datasets and high-capacity model architectures. In robotics\, however\, it has been challenging to apply the same recipe: after all\, we cannot easily scrape millions of hours of robot data from the internet and existing model architectures for scalable learning in vision or language modeling are not designed for the continuous control tasks we need to solve in robotics. In this talk\, I will describe my work on unlocking scalable robot learning in the real world. I will discuss important differences between robotics and other machine learning domains\, and describe how we can adapt scalable learning approaches for robotics to yield robot policies that generalize out-of-the-box to unseen environments\, and can be quickly adapted to new tasks or robot embodiments. Through a combination of community-wide data sharing efforts\, improved model design and training objectives\, my work has enabled the construction of the largest robot learning datasets to date\, and the training of generalist robot policies that can perform a range of complex\, long-horizon manipulation tasks simply by prompting them in natural language. I will close with a description of current limitations and open challenges towards building truly general robot control policies.
URL:https://seasevents.nmsdev7.com/event/cis-seminar-unlocking-scalable-robot-learning-in-the-real-world/
LOCATION:Levine 307\, 3330 Walnut Street\, Philadelphia\, PA\, 19104\, United States
ORGANIZER;CN="Computer and Information Science":MAILTO:cherylh@cis.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250328T093000
DTEND;TZID=America/New_York:20250328T103000
DTSTAMP:20260403T135429
CREATED:20250324T171232Z
LAST-MODIFIED:20250324T171232Z
UID:10008326-1743154200-1743157800@seasevents.nmsdev7.com
SUMMARY:CIS Seminar: "Learning to Perceive the 4D World"
DESCRIPTION:Perceiving the 4D world (i.e.\, 3D space over time) from visual input is essential for human interaction with the physical environment. While computer vision has made remarkable progress in 3D scene understanding\, much of it remains piecemeal—for example\, focusing solely on static scenes or specific categories of dynamic objects. How can we model diverse dynamic scenes in the wild? How can we achieve online perception with human-like capabilities? In this talk\, I will first discuss holistic scene representations that enable long-range motion estimation and 4D reconstruction. I will then introduce a unified learning-based framework for online dense 3D perception\, which continuously refines scene understanding with new observations. I will conclude by discussing future directions and challenges in advancing spatial intelligence.
URL:https://seasevents.nmsdev7.com/event/cis-seminar-learning-to-perceive-the-4d-world-2/
LOCATION:Amy Gutmann Hall\, Room 414\, 3333 Chestnut Street\, Philadelphia\, 19104\, United States
ORGANIZER;CN="Computer and Information Science":MAILTO:cherylh@cis.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250328T093000
DTEND;TZID=America/New_York:20250328T103000
DTSTAMP:20260403T135429
CREATED:20250327T203006Z
LAST-MODIFIED:20250327T203006Z
UID:10008332-1743154200-1743157800@seasevents.nmsdev7.com
SUMMARY:CIS Seminar: "Pareto-efficient AI systems: Expanding the quality and efficiency frontier of AI"
DESCRIPTION:We have made exciting progress in AI by massive models on massive amounts of data center compute. However\, the demands for AI are rapidly expanding. I identify how to maximize performance under any compute constraint\, expanding the Pareto frontier of AI capabilities.\n \nThis talk builds up to an efficient language model architecture that expands the Pareto-frontier between quality and throughput efficiency. In motivation\, the Transformer\, AI’s current workhorse architecture\, is memory hungry\, severely limiting its throughput\, or amount of text it can process per second. This has led to a Cambrian explosion of alternate efficient architecture candidates proposed across prior work. Prior work has painted an exciting picture: there exists architectures that are asymptotically faster than Transformers\, while also matching quality. However\, I ask\, if we’re using asymptotically faster building blocks\, are we giving something up in quality?\n\n\nIn part one\, we build understanding. Indeed\, there’s no free lunch! I present my work to identify and explain the fundamental quality and efficiency tradeoffs between different classes of architectures. Methods I developed for this analysis are now ubiquitous in the development of language models.\n\n\nIn part two\, we measure how AI architecture candidates fare on the tradeoff space. A major hurdle\, however\, is that we lack implementations of the architectures that that run at peak-efficiency on modern hardware. Further\, many proposed architectures are asymptotically fast\, but not wall-clock fast. I present ThunderKittens\, a new programming library I built to help AI researchers write simple\, hardware-efficient algorithms across hardware platforms.\n\n\nIn part three\, we expand the Pareto-frontier of the tradeoff space. I present the BASED architecture\, which is built from simple\, hardware-efficient components. I released the state-of-the-art 8B-405B Transformer-free language models\, per standard evaluations\, all on an academic budget.\n\n\nGiven the massive investment into language models\, this work has had significant impact and adoption in research\, open-source\, and industry.
URL:https://seasevents.nmsdev7.com/event/cis-seminar-pareto-efficient-ai-systems-expanding-the-quality-and-efficiency-frontier-of-ai-2/
LOCATION:Amy Gutmann Hall\, Room 414\, 3333 Chestnut Street\, Philadelphia\, 19104\, United States
ORGANIZER;CN="Computer and Information Science":MAILTO:cherylh@cis.upenn.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250328T103000
DTEND;TZID=America/New_York:20250328T114500
DTSTAMP:20260403T135429
CREATED:20241211T162547Z
LAST-MODIFIED:20241211T162547Z
UID:10008196-1743157800-1743162300@seasevents.nmsdev7.com
SUMMARY:Spring 2025 GRASP on Robotics: Mac Schwager\, Stanford University\, “Perception-Rich Robot Autonomy with Neural Environment Models”
DESCRIPTION:This will be a hybrid event with in-person attendance in Wu and Chen and virtual attendance on Zoom. \nABSTRACT\nRecent advances in computer vision have led to the rise of highly expressive 3D scene models such as NeRFs and GSplats. More than just rendering lifelike images\, these models allow robots to ground visual\, semantic\, physical\, and affordance properties in a common 3D model\, to rearrange objects in the scene and even simulate physical interactions. In this talk I will describe our efforts to build new robot autonomy features around these models\, while preserving safety\, modularity\, and interpretability. I will present navigation algorithms for robots to safely maneuver through their environment using NeRFs and GSplats\, even while training the model online in a SLAM-like fashion. I will describe methods to embed semantic and affordance information into radiance fields\, giving robots a 3D grounding for understanding and executing tasks from natural language commands. Finally I will describe using these neural models as high-fidelity training environments for learning end-to-end visuo-motor policies. I will demonstrate such a policy for navigating a drone through an obstacle-rich environment while being robust to significant visual distractors. I will conclude with future opportunities and challenges in neural environment models for robotics.
URL:https://seasevents.nmsdev7.com/event/spring-2025-grasp-on-robotics-mac-schwager-stanford-university-perception-rich-robot-autonomy-with-neural-environment-models/
LOCATION:Wu and Chen Auditorium (Room 101)\, Levine Hall\, 3330 Walnut Street\, Philadelphia\, PA\, 19104\, United States
ORGANIZER;CN="General Robotics%2C Automation%2C Sensing and Perception (GRASP) Lab":MAILTO:grasplab@seas.upenn.edu
END:VEVENT
END:VCALENDAR