http://talks.cam.ac.uk/show/rss/5408

Abstract not available

• Speaker: Seamus Branch
• Tuesday 13 May 2025, 13:10-14:00
• Venue: Richard King room, Darwin College.
• Series: Darwin College Humanities and Social Sciences Seminars; organiser: Matthew Jones.

Add to your calendar or Include in your list

Numerical evolutions show that, in spherical symmetry, as we move through the solution space of GR to the threshold of black hole formation, the resulting spacetimes tend to display a surprising degree of simplicity. A heuristic description of this behavior, called critical collapse, has been built around this empirical fact. Less is known when symmetry is dropped. In this presentation I will review the current status of the topic, focusing in particular on the struggle to understand the situation in axisymmetry.

• Speaker: David Hilditch, IST Lisbon
• Friday 02 May 2025, 13:00-14:00
• Venue: Potter room/Zoom https://cam-ac-uk.zoom.us/j/87235967698.
• Series: DAMTP Friday GR Seminar; organiser: Daniela Cors.

Add to your calendar or Include in your list

Taking the form of a guided tasting, this talk will explore some of the key questions around the science and history of cacao cultivation and chocolate production.

• Speaker: Mika Hyman, History and Philosophy of Science
• Thursday 01 May 2025, 13:00-14:00
• Venue: 1 Newnham Terrace, Darwin College.
• Series: Darwin College Science Seminars; organiser: Alexander R Epstein.

Add to your calendar or Include in your list

Machine learning potentials (MLPs) have become an important tool for atomistic simulations in many fields, from chemistry to materials science. The reason for the popularity of MLPs is their ability to provide very accurate energies and forces, which are essentially indistinguishable from the underlying reference electronic structure calculations. Still, the computational costs are much reduced enabling large-scale simulations of complex systems. Almost two decades ago, in 2007, the introduction of high-dimensional neural network potentials (HDNNP) by Behler and Parrinello paved the way for the application of MLPs to condensed systems containing a large number of atoms. Still, the original second-generation HDNN Ps, like most current MLPs, are based on a locality approximation of the atomic interactions that are truncated at some finite distance. Third-generation MLPs contain long-range electrostatic interactions up to infinite distance and overcome this restriction to short-range energies. Still, there are surprisingly many systems in which long-range electrostatic interactions are insufficient for a physically correct description, since non-local phenomena like long-range charge transfer are essential. Such global effects can be considered in fourth-generation HDNN Ps. In this talk the evolution of HDNN Ps will be discussed along with some key systems illustrating their applicability.

• Speaker: Prof. Jörg Behler (Ruhr-Universität Bochum)
• Thursday 19 June 2025, 14:00-15:30
• Venue: Seminar Room 3, RDC.
• Series: Theory of Condensed Matter; organiser: Bo Peng.

Add to your calendar or Include in your list

Axion dark matter is an interesting candidate for several reasons. Axions or axion-like particles appear in many theories beyond the standard model and there is a theoretical motivation for them to be light. They have many desirable properties, but also predict effects that are challenging for heavier, cold dark matter. I will discuss how there are observables connected to these effects that could be observed in the lab with a focus on quantum sensors, as well as novel approaches at colliders.

• Speaker: Martin Bauer: Durham University, IPPP
• Tuesday 29 April 2025, 11:00-12:00
• Venue: Seminar Room -- RDC D2.002 .
• Series: Cavendish HEP Seminars; organiser: Dr Paul Swallow.

Add to your calendar or Include in your list

Title

Machine Learning for Building-Level Heat Risk Mapping

Abstract

Climate change is intensifying the frequency and severity of heat waves, increasing risks to public health and energy systems worldwide. However, many existing heat vulnerability assessments focus primarily on outdoor temperatures, overlooking indoor conditions that directly affect occupants. Although building simulations can reveal the types of buildings whose occupants are most at risk, they rarely pinpoint the exact locations of these vulnerable buildings. In this presentation, I will present a data-driven workflow that locates high-risk buildings and discuss the labeling strategies we have explored for classifying real-world structures using satellite imagery.

Bio

Andrea is a first-year PhD student in the Department of Computer Science and Technology at the University of Cambridge. She is supervised by Prof Srinivasan Keshav. Her research bridges machine learning with civil and environmental engineering, focusing particularly on its applications within the built environment.

• Speaker: Andrea Domiter, University of Cambridge
• Thursday 08 May 2025, 13:00-14:00
• Venue: Room SS03 at the William Gates Building. Zoom link: https://cl-cam-ac-uk.zoom.us/j/4361570789?pwd=Nkl2T3ZLaTZwRm05bzRTOUUxY3Q4QT09&from=addon .
• Series: Energy and Environment Group, Department of CST; organiser: lyr24.

Add to your calendar or Include in your list

Taking the form of a guided tasting, this talk will explore some of the key questions around the science and history of cacao cultivation and chocolate production.

• Speaker: Mika Hyman, History and Philosophy of Science
• Thursday 01 May 2025, 13:00-14:00
• Venue: 1 Newnham Terrace, Darwin College.
• Series: Darwin College Science Seminars; organiser: Alexander R Epstein.

Add to your calendar or Include in your list

This talk explores the challenges faced by autonomous robots operating in human environments, focusing on our efforts to help them understand and clarify ambiguous instructions and act in socially appropriate ways while carrying out complex tasks.

• Speaker: Fethiye Irmak Dogan (University of Cambridge, Department of Computer Science and Technology)
• Thursday 08 May 2025, 13:00-14:00
• Venue: Richard King room, Darwin College.
• Series: Darwin College Science Seminars; organiser: Alexander R Epstein.

Add to your calendar or Include in your list

Most natural organisms show fascinating mechanical adaptability when interacting with their environments. Stiffness tuning in nature is used as a powerful tool to combine the load carrying functionality of rigid structures with compliance and adaptability. Human-made structures, however, do not possess this mechanical adaptability and are often designed to meet a specific load carrying requirement. This causes limitations in performance, efficiency and safety. Often to add other functionalities, additional components are needed, which increases the total weight and cost of the structures.

In this talk, I will present the latest research in our group on a variety of structures including multi-material cellular and multi-layered structures that employed active stiffness tuning based on thermoplastic softening. We use a combined experimental and numerical approach to investigate the electro-thermo-mechanical response of these structures. Understanding the main physical obstacles that limit the response time and the fundamental parameters controlling the stability and the failure under harsh electro-thermal loading will help us to better engineer the structures to meet the fast response and low power requirements. This new understanding will accelerate the technology readiness level of active structural control technology to be used in future multi-functional and smart structures. This technology has a wide range of application in robotics, morphing and deployable structures, active damping and active impact protection.

• Speaker: Iman Mohagheghian, University of Surrey
• Friday 02 May 2025, 15:00-16:00
• Venue: CivEng Seminar Room (1-33) (Civil Engineering Building).
• Series: Engineering Department Structures Research Seminars; organiser: Shehara Perera.

Add to your calendar or Include in your list

Over the last decade, we’ve seen unprecedented progress in AI across many disciplines and applications. However, autonomous vehicles are still far from mainstream even after billions of dollars of investment. In this talk we’ll explore what’s been holding progress back, and how by adopting a modern embodied AI approach to the problem, Wayve is finally unlocking the potential of scalable autonomous driving across the globe.

We’ll also explore some of our latest research in multimodal learning to combine the power of large language models with the driving problem, and in controllable generative world models as learned simulators.

Bio: Jamie leads Wayve’s Science department, where he guides our research teams to unlock new research breakthroughs, to enable those breakthroughs to have meaningful impact for the business, and to disrupt both our technical and business strategy to ensure Wayve stays at the forefront of innovation. Jamie has been at the forefront of applied AI research for the past 20 years. Before joining Wayve, Jamie was Partner Director of Science at Microsoft and Head of the Mixed Reality & AI Labs. While at Microsoft, Jamie shipped foundational features for Microsoft’s Kinect (Microsoft’s line of motion sensing input devices) and the hand- and eye-tracking that enable HoloLens 2’s interaction model (smart glasses). Jamie has a PhD in computer vision from the University of Cambridge and has received multiple Best Paper and Best Demo Awards at top-tier academic conferences. He was elected a Fellow of the Royal Academy of Engineering in 2021.

• Speaker: Jamie Shotton, Wayve
• Friday 02 May 2025, 14:00-15:00
• Venue: Computer Lab, LT1.
• Series: Cambridge ML Systems Seminar Series; organiser: Sally Matthews.

Add to your calendar or Include in your list

The Hosts for this talk are Nicky Clayton and Max Knowles

• Speaker: Peter Godfrey Smith
• Friday 30 May 2025, 16:30-18:00
• Venue: Ground Floor Lecture Theatre, Department of Psychology.
• Series: Zangwill Club; organiser: Sara Seddon.

Add to your calendar or Include in your list

The Host for this talk is Sarah-Jayne Blakemore

The host for this talk is Sarah-Jayne Blakemore

• Speaker: Professor Ron Mangun,Center for Mind and Brain 267 Cousteau Place Davis, CA
• Friday 23 May 2025, 16:30-18:00
• Venue: Ground Floor Lecture Theatre, Department of Psychology.
• Series: Zangwill Club; organiser: Sara Seddon.

Add to your calendar or Include in your list

The host for this talk is Nicky Clayton

Abstract: The talk will present research from our research team where we have explored how the structure of the environment affects wayfinding behaviour. It will cover our research with Sea Hero Quest in which we found growing up in griddy cities has a negative impact on navigation behaviour, as well as well as research with London taxi drivers how the environment affects how they plan. In the second part I will cover our recent research in neuroarchitecture exploring brain responses (fMRI) during watching movies of pleasant or unpleasant built environment and crowd dynamics in a study of 100 people navigating and exploring a fabricated large-scale art gallery (The 100 Minds in Motion Project).

Bio: Hugo Spiers is Professor of Cognitive Neuroscience, and a Vice Dean for Enterprise, at University College London (UCL). He has over 25 years of research experience in neuroscience and psychology studying how our brain recalls the past, navigates the present and imagines the future. He has published over 100 academic articles and received numerous awards including the Charles Darwin Award from the British Science Association and a James McDonnell Foundation Scholar Award. He is co-director of the International Centre for NeuroArchitecture and NeuroDesign, a Fellow of the Royal Institute of Navigation, a Lighthouse Fellow of the Centre for Conscious Design and the Vice Chair of the Academy of Neuroscience for Architecture in the UK. His research project Sea Hero Quest has tested over 4 million people in 195 nations on their navigation ability, providing a powerful benchmark for assessment in Alzheimer’s disease and global insight into cognition.

• Speaker: Professor Hugo Spiers
• Friday 02 May 2025, 16:30-18:00
• Venue: Ground Floor Lecture Theatre, Department of Psychology.
• Series: Zangwill Club; organiser: Sara Seddon.

Add to your calendar or Include in your list

Data-driven modeling typically involves simplifications of systems through dimensionality reduction (less variables) or through dimensionality enlargement (more variables, but simpler, perhaps linear, dynamics). Autoencoders with narrow bottleneck layers are a typical approach to the former (allowing the discovery of dynamics taking place in a lower-dimensional manifold), while autoencoders with wide layers provide an approach to the later, with “neurons” in these layers thought of as “observables” in Koopman representations. In the first part of this talk, I’ll briefly discuss some theoretical results about each of these topics. (Joint work with M.D. Kvalheim on dimension reduction and with Z. Liu and N. Ozay on Koopman representations.)

The training of autoencoders, and more generally the solution of other optimization problems, including policy optimization in reinforcement learning, typically relies upon some variant of gradient descent. There has been much recent work in the machine learning, control, and optimization communities in the application of the Polyak-Łojasiewicz Inequality (PŁI) to such problems in order to establish exponential (a.k.a. “linear” in the local-iteration language of numerical analysis) convergence of loss functions to their minima under the gradient flow. A somewhat surprising fact is that the exponential rate, at least in the continuous-time LQR problem, vanishes for large initial conditions, resulting in a mixed globally linear / locally exponential behavior. This is in sharp contrast with the discrete-time LQR problem, where there is global exponential convergence. The gap between CT and DT behaviors motivated our work on generalizations of the PŁI condition, and the second part of the talk will address that topic. In fact, these generalizations are key to understanding the effect of errors in the estimation of the gradient. Such errors might arise from adversarial attacks, wrong evaluation by an oracle, early stopping of a simulation, inaccurate and very approximate digital twins, stochastic computations (algorithm “reproducibility”), or learning by sampling from limited data. We will suggest an input to state stability (ISS) analysis of this issue. Time permitting, we will also mention some initial results on the performance of linear feedforward networks in feedback control. (Joint work with A.C.B. de Oliveira, L. Cui, Z.P. Jiang, and M. Siami).

The seminar will be held in JDB Seminar Room , Department of Engineering, and online (zoom): https://newnham.zoom.us/j/92544958528?pwd=YS9PcGRnbXBOcStBdStNb3E0SHN1UT09

• Speaker: Eduardo Sontag, Northeastern University
• Thursday 01 May 2025, 14:00-15:00
• Venue: JDB Seminar Room, Department of Engineering and online (Zoom).
• Series: CUED Control Group Seminars; organiser: Fulvio Forni.

Add to your calendar or Include in your list

Over the last decade, we’ve seen unprecedented progress in AI across many disciplines and applications. However, autonomous vehicles are still far from mainstream even after billions of dollars of investment. In this talk we’ll explore what’s been holding progress back, and how by adopting a modern embodied AI approach to the problem, Wayve is finally unlocking the potential of scalable autonomous driving across the globe.

We’ll also explore some of our latest research in multimodal learning to combine the power of large language models with the driving problem, and in controllable generative world models as learned simulators.

Bio: Jamie leads Wayve’s Science department, where he guides our research teams to unlock new research breakthroughs, to enable those breakthroughs to have meaningful impact for the business, and to disrupt both our technical and business strategy to ensure Wayve stays at the forefront of innovation. Jamie has been at the forefront of applied AI research for the past 20 years. Before joining Wayve, Jamie was Partner Director of Science at Microsoft and Head of the Mixed Reality & AI Labs. While at Microsoft, Jamie shipped foundational features for Microsoft’s Kinect (Microsoft’s line of motion sensing input devices) and the hand- and eye-tracking that enable HoloLens 2’s interaction model (smart glasses). Jamie has a PhD in computer vision from the University of Cambridge and has received multiple Best Paper and Best Demo Awards at top-tier academic conferences. He was elected a Fellow of the Royal Academy of Engineering in 2021.

• Speaker: Jamie Shotton, Wayyve
• Friday 02 May 2025, 14:00-15:00
• Venue: Computer Lab, LT1.
• Series: Cambridge ML Systems Seminar Series; organiser: Sally Matthews.

Add to your calendar or Include in your list

Abstract not available

• Speaker: Szilvia Ujvary (University of Cambridge)
• Friday 20 June 2025, 12:00-13:00
• Venue: ONLINE ONLY. Here is the Zoom link: https://cam-ac-uk.zoom.us/j/4751389294?pwd=Z2ZOSDk0eG1wZldVWG1GVVhrTzFIZz09.
• Series: NLIP Seminar Series; organiser: Suchir Salhan.

Add to your calendar or Include in your list

In this talk, I will show the representation power of sphere embedding: (1) a variety of basic geometric objects can be defined as spheres; (2) Spheres are perfect for neuro-symbolic unification – sphere centre hosting latent feature embedding for associative thinking (System 1), sphere boundary explicitly embodies part-whole or category relations for logical reasoning (System 2); (3) Spheres and the connection relation can define commonsense spatial relations, including orientation and distance; (4) Cartesian product of sphere configurations can formulate event structure. Putting all these together, we can use sphere configurations to define the knowledge structure of humour, which is not far away from the knowledge structure of syllogism. This talk ends with the conclusion: spheres are natural representations of syllogistic reasoning (the microcosm of human rationality) and humour reasoning (the highest level of human rationality). They are the two end-poles that span the continuum of human intelligent behaviours.

• Speaker: Tiansi Dong
• Tuesday 29 April 2025, 17:00-17:45
• Venue: Lecture Theatre 2, Computer Laboratory, William Gates Building.
• Series: Foundation AI; organiser: Pietro Lio.

Add to your calendar or Include in your list

Abstract not available

• Speaker: Tiansi Dong
• Saturday 26 April 2025, 17:00-17:45
• Venue: Lecture Theatre 2, Computer Laboratory, William Gates Building.
• Series: Foundation AI; organiser: Pietro Lio.

Add to your calendar or Include in your list

Abstract not available

• Speaker: Tiansi Dong
• Saturday 26 April 2025, 17:00-17:45
• Venue: Lecture Theatre 2, Computer Laboratory, William Gates Building.
• Series: Foundation AI; organiser: Pietro Lio.

Add to your calendar or Include in your list

Abstract not available

• Speaker: Tiansi Dong
• Saturday 26 April 2025, 17:00-17:45
• Venue: Lecture Theatre 2, Computer Laboratory, William Gates Building.
• Series: Foundation AI; organiser: Pietro Lio.

Add to your calendar or Include in your list