Glasses are generally regarded as disordered and the idea of “controlling” molecular packing in glasses is reasonably met with skepticism. However, as glasses are non-equilibrium materials, a vast array of amorphous structures are possible in principle. Physical vapor deposition (PVD) produces glasses with properties that cannot be achieved by other preparation routes, including high stability […]
Soft robots have the potential to adapt their morphologies and behavioral control policies to changing tasks and environments. Inspired by the dynamic plasticity of living organisms and the general adaptability of animals, this talk will discuss several shape-shifting soft robot platforms for multi-task performance and multi-environment locomotion—for example, robotic skins, robotic fabrics, and robots with […]
In classical auction design, we take it for granted that the auctioneer is trusted and always implements the auction's rules honestly. This assumption, however, no longer holds in modern auctions based on blockchains, or those mediated by third-party platforms such as Google. For example, in blockchain-based auctions, the consensus nodes that partly act as the […]
Abstract: The efficient sharing of AI infrastructures is becoming increasingly important in both public and private data centers. This demand is driven by two key factors: the proliferation of specialized AI models tailored for different users and applications, and the highly dynamic nature of requests, which are often on-demand. Dedicated GPU allocation in such scenarios […]
This will be a hybrid event with in-person attendance in Levine 307 and virtual attendance on Zoom. ABSTRACT Modern artificial intelligences (AIs) rely heavily on internet-scale data with unified representations. However, such large-scale homogeneous data isn't readily available for spatial computing applications involving 3D geometry, hindering the development of spatial intelligence— AIs that can generate […]
Abstract: Neurological conditions are the leading cause of illness worldwide, though over 92% of clinically tested CNS drug candidates fail to become treatments. Contributing to this high failure rate is a lack of understanding of human disease mechanisms, technologies to address them, and the restrictive blood-brain barrier (BBB), which most compounds fail to cross. New […]
Metals powders like aluminum and boron are attractive potential fuel additives for pyrotechnics, propellants and explosives due to their high energy release upon oxidation. However, they tend to agglomerate, have lengthy ignition delays, and low combustion rates/efficiencies. This work aims to design metal powders with tuned surface, micro-structure, morphology, or chemistries to mitigate these challenges […]
Our exponential technologies for reading, writing genomes and epigenomes combined with AI-ML has enabled large libraries and selections for radical new functions — e.g. resistance to all viruses, novel delivery vectors for gene therapies, xeno-transplantation, de-extinction, and de-aging. This lecture will be held remotely. Zoom Link: https://upenn.zoom.us/j/94497096918?pwd=5qeoyMcmEyDboMW3bKgSCmWKYBMCbP.1
This will be a hybrid event with in-person attendance in Wu and Chen and virtual attendance on Zoom. ABSTRACT Motivated by the ever-increasing success of machine learning in language and vision models, many aim to build AI-driven tools for scientific simulation and discovery. Contemporary techniques drastically lag behind their comparatively mature counterparts in modeling and […]
Reliable forecasts of the Earth system are crucial for human progress and safety from natural disasters. Artificial intelligence offers substantial potential to improve prediction accuracy and computational efficiency in this field, however this remains underexplored in many domains. Here we introduce Aurora, a large-scale foundation model for the Earth system trained on over a million […]
Abstract: Polymer networks, including thermoplastics, thermosets, elastomers and gels, are among the most versatile and widely utilized polymeric materials, with applications spanning drug delivery systems, membranes and implantable devices. A comprehensive understanding of the intricate relationship between the macroscopic properties and the structure and topology of polymer networks is crucial for advancing their utilization and […]
Scalable and efficient machine learning (ML) systems have been instrumental in fueling recent advancements in ML capabilities. However, further scaling these systems requires more than simply increasing the number and performance of accelerators. This is because modern ML deployments rely on complex pipelines composed of many diverse and interconnected systems. In this talk, I will […]
Abstract: Modern machine learning models—such as LLMs and recommender systems—interact with humans, companies, and other models in a broader ecosystem. However, these multi-agent interactions often induce unintended ecosystem-level outcomes such as clickbait in classical content recommendation ecosystems, and more recently, safety violations and market concentration in nascent LLM ecosystems. In this talk, I discuss my […]
This will be a hybrid event with in-person attendance in Towne 100 Heilmeier Hall and virtual attendance on Zoom. ABSTRACT In this talk, I present an approach for interpreting the internal computation in deep vision models. I show that these interpretations can be used to detect model bugs and to improve the performance of pre-trained […]
Abstract: The discovery of functional molecules plays a fundamental role in advancing chemical science and engineering, yet it remains a costly and time-intensive process. Recent advances in computational methods, particularly in generative artificial intelligence, have introduced a new approach—generative molecular design—which holds the promise of efficiently identifying molecules with desired properties. However, despite significant progress, […]
Cells across tissues, developmental stages, and disease conditions adopt distinct intracellular states – epigenomic, transcriptomic, and proteomic profiles – to compartmentalize function in time and space. The ability to write DNA- and RNA-encoded programs that sense and interface with cellular states has transformative potential for biotechnology, for example in designing gene therapies with cell type- […]
This will be a hybrid event with in-person attendance in Wu and Chen and virtual attendance on Zoom. ABSTRACT Control tuning and adaptation present a significant challenge to the usage of robots in diverse environments. It is often nontrivial to find a single set of control parameters by hand that work well across the broad […]
The growing global energy demand has driven the development of advanced batteries, particularly lithium metal anodes and anode-free systems, due to their potential for higher energy density at reduced costs. However, characterizing the interfaces within these systems presents a critical challenge due to their susceptibility to decomposition during conventional atomic-resolution analysis. The emergence of Cryogenic […]
Structural failure is a critically important design consideration in many engineering applications. Nature provides a number of interesting examples of lightweight structural features that exhibit outstanding failure characteristics. Recent progress in additive manufacturing has facilitated precise control over geometric features, allowing for the implementation of bio-inspired structural designs on a layer-by-layer basis. In this thesis, […]
"Creating Lasting Impact at the Interface of Energy, Technology, and Society" Join Vijay Kumar, Nemirovsky Family Dean of Penn Engineering for a discussion with David Turk – the former Deputy Secretary of the U.S. Department of Energy and Vanessa Chan, the former Chief Commercialization Officer and Director of the Office of Technology Transitions at the […]
Flow systems in nature, whether animate or inanimate, have evolved in time toward greater efficiency by adapting, or ‘morphing’, their configuration to decrease resistance to the currents flowing through them. Observations at various scales indicate that the distribution of flow, i.e. the connection of a point (source/sink) to a volume (sink/source) or vice-versa, is the […]
Algorithms are increasingly integrated into various societal applications, often directly interacting with people and communities. This highlights the importance of understanding the interplay between algorithmic decisions and economic incentives when designing machine learning algorithms. In this talk, I will explore two examples of this dynamic through the lens of privacy in data markets and fairness […]
In this panel, senior members of the Biden-Harris Administration will discuss the work they did to catalyze the private sector to commercialize critical technologies for the clean energy transition. Through the Bipartisan Infrastructure Law and the Inflation Reduction Act, the U.S. Department of Energy (DOE) created over 900,000 new jobs, catalyzed over 900 new and […]
Abstract: Large Language Models (LMs) have demonstrated remarkable capabilities by scaling up training data and model sizes. However, they continue to face critical challenges, including hallucinations and outdated knowledge, which particularly limit their reliability in expert domains such as scientific research and software development. In this talk, I will urge the necessity of moving beyond […]
Abstract: Nature harnesses genetic sequences to encode structural information and emergent physical and chemical transformations required by life processes. Such orchestration is achieved through hierarchical self-assemblies driven by weak molecular interactions and optimized via evolution. However, emulating this level of programmability to create designer biomolecular materials remains a grand challenge. In this seminar, I will […]
Vanessa Chan, the Jonathan and Linda Brassington Practice Professor in Materials Science and Engineering (MSE) and Vice Dean of Innovation and Entrepreneurship at Penn Engineering, is hosting the premiere of the documentary Legion 44. Special guests include Leila Conners, the film’s director and producer, and Dr. Max Pisciotta, a featured climate scientist in the film […]
Abstract: Terrestrial plants harbor renewable carbon resources in their cell walls that can directly contribute to decarbonization of the transportation and materials sectors, but the heterogeneous, aromatic lignin polymer found in plants has long obstructed both natural- and humankind-driven endeavors to deconstruct plant biomass to valuable products. However, viable methods to valorize lignin are essential […]
The Environmental Innovations Initiative's Climate 101 series demystifies climate change and its link with a diversity of disciplines. Experts from across Penn share how they think about climate through the lens of their research and teaching, helping to raise our climate literacy and prepare us to tackle one of the planet's greatest challenges: the climate […]
Deep learning's success stems from the ability of neural networks to automatically discover meaningful representations from raw data. In this talk, I will describe some recent insights into how optimization enables this learning process. First, I will show how optimization algorithms exhibit surprisingly rich dynamics when training neural networks, and how these complex dynamics are […]
To meet net-zero carbon emissions targets by mid-century, up to a -fold increase in wind power capacity is required. Acceleration to this rate requires urgent improvements to efficiency and reliability of installed wind farms, as well as cost reductions for future offshore farms. To expand energy production, wind turbines are rapidly increasing in size, wind […]
Designing miniaturized robots and bioelectronic devices will enable access throughout the entire human body, leading to novel procedures at the cellular level and offering localized diagnosis and treatment with unprecedented precision and efficiency. However, the soft, complex, and multi-dimensional nature of biological systems poses significant challenges for mechanical design, manufacturing, materials engineering, and functional integration […]
Conventional approaches to scientific discovery often prioritize building larger sensors, gathering more data, and scaling up computational power. In this talk, I will present a complementary perspective: extracting insights hidden in the data we already have. The key lies in using AI not as a black-box predictor, but as a tool for interpreting data through […]
The PCP theorem, proved in the 90’s, shows how to encode a proof for any theorem into a format where the theorem's correctness can be verified by making only a constant number of queries to the proof. This result is a significant milestone in computer science and has important implications for approximation algorithms, cryptography, and […]
Recent advancements in hardware and software are bringing autonomous aerial vehicles closer than ever to finally delivering on futuristic visions of flying cars and package delivery drones. However, the safe deployment of autonomous aircraft at scale in urban environments poses significant challenges, one of which being uncertainties contributed by complex spatial and temporal winds. Clusters […]
Abstract: Large language models (LLMs) power a rapidly-growing and increasingly impactful suite of AI technologies. However, due to their scale and complexity, we lack a fundamental scientific understanding of much of LLMs’ behavior, even when they are open source. The “black-box” nature of LMs not only complicates model debugging and evaluation, but also limits trust […]
This will be a hybrid event with in-person attendance in Levine 307 and virtual attendance on Zoom. ABSTRACT Partial observability is ubiquitous in Reinforcement Learning (RL) applications, where agents must make sequential decisions despite lacking complete information about the latent states of the controlled system. Partially observable RL is notoriously challenging in theory—well-known information-theoretic results […]
Arrays of coupled superconducting qubits are a compelling platform for analog quantum simulations of solid-state matter and many-body physics. These devices natively emulate the Bose-Hubbard model while offering a high degree of control, fast operation rates, and site-resolved readout. We discuss recent experiments using a 4-by-4 array of transmon qubits. By adopting a parametric coupling […]
As learning algorithms become increasingly capable of acting autonomously, it is important to better understand the behavior that results from their interactions. For example, a pervasive challenge in multi-agent learning settings, which spans both theory and practice and dates back decades, has been the failure of convergence for iterative algorithms such as gradient descent. Accordingly, […]
In this talk, I will explain how fundamental problems in computational learning theory are at the heart of modern problems in machine learning and scientific applications and how algorithmic insights in mathematically tractable models can inspire new solutions in a wide variety of domains. I will explore two directions. First, I will explore algorithmic foundations […]
This will be a hybrid event with in-person attendance in Wu and Chen and virtual attendance on Zoom. This seminar will NOT be recorded. ABSTRACT State of the art prediction can fail catastrophically when operating outside of the training distribution. We show methods that can be used to reduce these failures. We then outline how […]
Understanding the dynamic behavior of complex biomolecules requires simplified models that not only make computations feasible but also reveal fundamental mechanisms. Coarse-graining (CG) achieves this by grouping atoms into beads, whose stochastic dynamics can be derived using the Mori-Zwanzig formalism, capturing both reversible and irreversible interactions. In liquid, the dissipative bead-bead interactions have so far […]
Robotic systems, particularly at small scales, require efficient actuation and sensing solutions that maintain compactness. We are interested in systems where sensing and actuation are seamlessly integrated, specifically using impedance-related properties—such as electrical resistance, induced electromotive force (emf), and inductance— for free self-sensing in actuators without additional sensors. We explore three main example applications: (1) […]
This will be an in-person event only with attendance in Levine 307. This seminar will NOT be recorded. ABSTRACT Autonomous AI agents are deployed in increasingly complex and uncertain environments where they must account for the presence of other agents while trying to achieve their own objectives. Moreover, such agents may require assistance from other […]
Accurate simulations of physical phenomena governed by partial differential equations (PDEs) are foundational to scientific computing. While traditional numerical methods have proven effective, they remain computationally intensive, particularly for complex, large-scale systems. This talk introduces the neural operator, a machine learning framework that approximates solution operators in infinite-dimensional spaces, enabling efficient and scalable PDE simulations […]
This will be a hybrid event with in-person attendance in Levine 512 and virtual attendance on Zoom. ABSTRACT Our group's goal is to build a world simulator from visual observations that can answer "what-if" questions. In pursuit of this goal, we develop various methods for modeling the world from images and harnessing physical simulation and […]
Modern computing and communication technologies, such as supercomputers and the internet, are based on optically-linked networks of information processors operating at microwave frequencies. An analogous architecture has been proposed for quantum networks using optical photons to distribute entanglement between remote superconducting quantum processors. Here I will discuss our recent demonstration of a chip-scale source of […]
This will be a hybrid event with in-person attendance in Levine 307 and virtual attendance on Zoom. ABSTRACT Construction remains one of the least automated industries, struggling with labor shortages, inefficiencies, and rising costs. At BotBuilt, we are revolutionizing the way homes are built by leveraging robotics and AI to automate framing, making construction faster, […]
Abstract: Societies have prospered using a linear “take-make-use-dispose” approach, extracting natural resources to make products, using them, and ultimately discarding them or their residues. This unsustainable approach has exploited natural resources at a rate that has caused excessive pollution and loss of biodiversity, and is leading to a global climate crisis. In response to this […]
Abstract: Fast solid-state Li-ion conductors are a crucial class of materials with the potential to enable all-solid-state batteries, offering enhanced safety and energy density. However, these materials remain rare, and progress in developing novel solid electrolytes has been hindered by a lack of clear descriptors for superionic conductivity and a limited understanding of ion transport […]
The remarkable recent success of Machine Learning (ML) is driven by our ability to develop and deploy interactive models that can solve complicated tasks by understanding and adapting to the ever-changing state of the world. However, the development of such models demands significant data and computing resources. Moreover, as these models increasingly interact with humans, […]
Bone is the third most common site for cancer metastasis, affecting ~66% of patients with common cancers—breast, lung, prostate, renal, thyroid—incurring skeletal events in up to 400,000 people in the US/year. Metastatic bone disease (MBD) results in weakened bone, leading to refractory pain and pathological fracture that increase disease state morbidity. Despite bone tissue’s dynamic […]
The rapid adoption of generative AI has created a cycle where personal information cascades perpetually: from people to models to applications and online platforms, then back through scrapers into the system. Simple blanket rules such as "don't train on this data" or "don't share sensitive information" are inadequate, as we face training data scarcity while […]
This will be a hybrid event with in-person attendance in Wu and Chen and virtual attendance on Zoom. ABSTRACT The goal of building machines that can perceive and act in the world as humans and other animals do has been a focus of AI research efforts for over half a century. Over this same period, […]
Special location for this talk: 105 Amy Gutmann Hall Recent advances in Artificial Intelligence are powering revolutionary interactive tools that will transform the very nature of the scientific enterprise. We describe several large-scale projects at the Allen Institute for AI aimed at developing open models, agentic platforms, and novel interaction paradigms in order to amplify […]