skip to content

Microstructural Kinetics Group

Department of Materials Science & Metallurgy
 

Tue 18 Feb 14:00: TBA

School of Physical Sciences - Mon, 03/02/2025 - 12:21
TBA

Abstract not available

  • Speaker: Titus Lupu (Paris)
  • Tuesday 18 February 2025, 14:00-15:00
  • Venue: MR12.
  • Series: Probability; organiser: ww295.

Add to your calendar or Include in your list

Fri 14 Mar 14:00: Evaluating a black-box algorithm: stability, risk, and model comparisons

School of Physical Sciences - Mon, 03/02/2025 - 12:16
Evaluating a black-box algorithm: stability, risk, and model comparisons

When we run a complex algorithm on real data, it is standard to use a holdout set, or a cross-validation strategy, to evaluate its behavior and performance. When we do so, are we learning information about the algorithm itself, or only about the particular fitted model(s) that this particular data set produced? In this talk, we will establish fundamental hardness results on the problem of empirically evaluating properties of a black-box algorithm, such as its stability and its average risk, in the distribution-free setting. This work is joint with Yuetian Luo and Byol Kim.

Add to your calendar or Include in your list

Thu 13 Mar 16:00: Algorithmic stability for regression and classification

School of Physical Sciences - Mon, 03/02/2025 - 11:59
Algorithmic stability for regression and classification

In a supervised learning setting, a model fitting algorithm is unstable if small perturbations to the input (the training data) can often lead to large perturbations in the output (say, predictions returned by the fitted model). Algorithmic stability is a desirable property with many important implications such as generalization and robustness, but testing the stability property empirically is known to be impossible in the setting of complex black-box models. In this work, we establish that bagging any black-box regression algorithm automatically ensures that stability holds, with no assumptions on the algorithm or the data. Furthermore, we construct a new framework for defining stability in the context of classification, and show that using bagging to estimate our uncertainty about the output label will again allow stability guarantees for any black-box model. This work is joint with Jake Soloff and Rebecca Willett.

Evaluating a black-box algorithm: stability, risk, and model comparisons

When we run a complex algorithm on real data, it is standard to use a holdout set, or a cross-validation strategy, to evaluate its behavior and performance. When we do so, are we learning information about the algorithm itself, or only about the particular fitted model(s) that this particular data set produced? In this talk, we will establish fundamental hardness results on the problem of empirically evaluating properties of a black-box algorithm, such as its stability and its average risk, in the distribution-free setting. This work is joint with Yuetian Luo and Byol Kim.

A wine reception in the Central Core will follow this lecture

Add to your calendar or Include in your list

Thu 13 Mar 16:00: Algorithmic stability for regression and classification

School of Physical Sciences - Mon, 03/02/2025 - 11:25
Algorithmic stability for regression and classification

In a supervised learning setting, a model fitting algorithm is unstable if small perturbations to the input (the training data) can often lead to large perturbations in the output (say, predictions returned by the fitted model). Algorithmic stability is a desirable property with many important implications such as generalization and robustness, but testing the stability property empirically is known to be impossible in the setting of complex black-box models. In this work, we establish that bagging any black-box regression algorithm automatically ensures that stability holds, with no assumptions on the algorithm or the data. Furthermore, we construct a new framework for defining stability in the context of classification, and show that using bagging to estimate our uncertainty about the output label will again allow stability guarantees for any black-box model. This work is joint with Jake Soloff and Rebecca Willett.

Evaluating a black-box algorithm: stability, risk, and model comparisons

When we run a complex algorithm on real data, it is standard to use a holdout set, or a cross-validation strategy, to evaluate its behavior and performance. When we do so, are we learning information about the algorithm itself, or only about the particular fitted model(s) that this particular data set produced? In this talk, we will establish fundamental hardness results on the problem of empirically evaluating properties of a black-box algorithm, such as its stability and its average risk, in the distribution-free setting. This work is joint with Yuetian Luo and Byol Kim.

Add to your calendar or Include in your list

Mon 10 Feb 14:00: Extreme pushed and pulled fronts

School of Physical Sciences - Mon, 03/02/2025 - 10:59
Extreme pushed and pulled fronts

I shall describe the propagation properties of a class of quasilinear reaction-diffusion equations, motivated by applications to biological tissue growth.

Add to your calendar or Include in your list

Tue 04 Feb 11:00: Planetary uprising: Climate colonialism, Extinction Rebellion and the transformation of global politics Teams link: https://teams.microsoft.com/l/meetup-join/19%3ameeting_Y2VlZmM3OTgtOTQwNS00ZTcxLTk5ZGEtZWZiMzU4NTdiMGY1%40thread.v2/0...

School of Physical Sciences - Mon, 03/02/2025 - 10:42
Planetary uprising: Climate colonialism, Extinction Rebellion and the transformation of global politics

Dear all,

CAS seminar will welcome Tobias Müller who will give us a talk on climate colonialism. The talk will be held in a hybrid format with the speaker in-person at the Unilever lecture theatre and on Zoom on Tuesday, the 4th February , 11 AM-12 PM. Please find the abstracts of the talk below.

If you would like this seminar recorded, please let us know in advance. We look forward to seeing you there!

Best wishes, Megan and Yao

—————————————

Leverhulme Early Career Fellow Centre for Research in the Arts, Social Sciences and Humanities (CRASSH) The climate crisis is deeply entangled with the politics of race and colonialism. The concept of “climate colonialism”, (Bhambra and Newell 2022) urges us to analyse what forms of resistance to the socio-ecological continuities of colonialism emerge, and what challenges they face. However, we lack empirical and conceptual studies on how people on the ground confront the intersection of the climate crisis, colonialism, racism and extractivism, and how this differs across former coloniser and colonised countries. This raises the question, what kind of politics are able to confront the intersecting crises of climate and colonialism?

This presentation addresses this gap through an analysis of how climate activists in four different countries respond to the climate crisis and connected social justice issues. Using the case study of a transnationally operating group within the global movement, Extinction Rebellion, the paper compares strategic responses to climate colonialism in four different countries, namely Mexico, South Africa, the UK and the US. Methodologically, the paper uses multi-sited ethnography, comprising 18 months of ethnographic fieldwork and 140 interviews with activists, to gain a deep insight into the internal contentions within different parts of the movement.The paper advances not only our understanding of how facing climate colonialism challenges movement spaces, but also how white environmental activists struggle with building racial justice into their practices and to build coalitions across the social justice movement space. It thereby contributes to the much-needed bridging between decolonial theory, social movement studies and the social scientific accounts of climate change.

Teams link: https://teams.microsoft.com/l/meetup-join/19%3ameeting_Y2VlZmM3OTgtOTQwNS00ZTcxLTk5ZGEtZWZiMzU4NTdiMGY1%40thread.v2/0?context=%7b%22Tid%22%3a%2249a50445-bdfa-4b79-ade3-547b4f3986e9%22%2c%22Oid%22%3a%2253b919d9-f8a7-4f56-9bb0-baaf0ba7404d%22%7d

Add to your calendar or Include in your list

Tue 04 Feb 14:30: Squarefree values of discriminant polynomials

School of Physical Sciences - Mon, 03/02/2025 - 10:29
Squarefree values of discriminant polynomials

Given a multivariable polynomial with integer coefficients, what is the probability that it takes a squarefree value? This was determined by Granville and Poonen assuming the ABC conjecture, but an unconditional proof remains unknown. We will start the talk by discussing a paper by Bhargava, Shankar and Wang that unconditionally determines a result for the discriminants of monic polynomials of any given degree. In the second part of the talk, we will see how the methods of BSW can be interpreted into a more general framework developed by Thorne, and how they can be used to obtain new results for other families of discriminant polynomials.

Add to your calendar or Include in your list

Wed 26 Feb 15:30: Title to be confirmed

School of Physical Sciences - Mon, 03/02/2025 - 09:27
Title to be confirmed

Abstract not available

Add to your calendar or Include in your list

Wed 23 Apr 14:00: Title to be confirmed

School of Physical Sciences - Mon, 03/02/2025 - 09:25
Title to be confirmed

Abstract not available

Add to your calendar or Include in your list

Wed 09 Apr 14:00: Title to be confirmed

School of Physical Sciences - Mon, 03/02/2025 - 09:24
Title to be confirmed

Abstract not available

Add to your calendar or Include in your list

Wed 26 Mar 14:00: Title to be confirmed

School of Physical Sciences - Mon, 03/02/2025 - 09:24
Title to be confirmed

Abstract not available

Add to your calendar or Include in your list

Wed 12 Mar 14:00: Title to be confirmed

School of Physical Sciences - Mon, 03/02/2025 - 09:23
Title to be confirmed

Abstract not available

Add to your calendar or Include in your list

Thu 15 May 14:00: Title to be confirmed

School of Physical Sciences - Mon, 03/02/2025 - 08:58
Title to be confirmed

Abstract not available

Add to your calendar or Include in your list

Fri 14 Feb 16:00: Synchronization in Navier-Stokes turbulence and it's role in data-driven modeling

School of Physical Sciences - Sat, 01/02/2025 - 10:27
Synchronization in Navier-Stokes turbulence and it's role in data-driven modeling

n Navier-Stokes (NS) turbulence, large-scale turbulent flows determine small-scale flows; in other words, small-scale flows are synchronized to large-scale flows. In 3D turbulence, previous numerical studies suggest that the critical length separating these two scales is determined by the Kolmogorov length. In this talk, I will introduce our theoretical framework for characterizing synchronization phenomena [1]. Specifically, it provides a computational method for the exponential rate of convergence to the synchronized state, and identifies the critical length based on the NS equations via the “transverse” Lyapunov exponent. I will also discuss the synchronization property of 2D NS turbulence and how it differs from the 3D case [2]. These insights into synchronization and critical length scales are essential for developing machine-learning closure models for turbulence, in particular their stable reproducibility [3]. Finally, I will illustrate how “generalized” synchronization is crucial for predicting chaotic dynamics [4].

[1] M. Inubushi, Y. Saiki, M. U. Kobayashi, and S. Goto, Characterizing small-scale dynamics of Navier-Stokes turbulence with transverse Lyapunov exponents: A data assimilation approach, Phys. Rev. Lett. 131, 254001 (2023).

[2] M. Inubushi and C. P. Caulfield (in preparation).

[3] S. Matsumoto, M. Inubushi, and S. Goto, Stable reproducibility of turbulence dynamics by machine learning, Phys. Rev. Fluids 9, 104601 (2024).

[4] A. Ohkubo and M. Inubushi, Reservoir computing with generalized readout based on generalized synchronization, Sci. Rep. 14, 30918 (2024).

Add to your calendar or Include in your list

Thu 05 Jun 14:00: Title to be confirmed

School of Physical Sciences - Thu, 30/01/2025 - 17:50
Title to be confirmed

Abstract not available

Add to your calendar or Include in your list

Fri 07 Feb 13:00: Unimodular JT gravity and de Sitter quantum cosmology

School of Physical Sciences - Thu, 30/01/2025 - 11:42
Unimodular JT gravity and de Sitter quantum cosmology

In this talk, I will show how a gauge-theoretic approach to Jackiw–Teitelboim (JT) gravity naturally yields a two-dimensional Henneaux–Teitelboim (HT) unimodular theory, applicable to both flat and curved spacetimes. Under a mini-superspace reduction, the Wheeler–DeWitt equation becomes a Schrödinger-like equation admitting a consistent, unitary quantum description. The resulting wavefunction describes a quantum distribution for the scale factor, illuminating cosmic expansion and contraction, and allowing topology change at a=0.

Add to your calendar or Include in your list

Mon 16 Jun 12:30: QBS

School of Physical Sciences - Wed, 29/01/2025 - 16:44
QBS

Abstract not available

Add to your calendar or Include in your list

Mon 19 May 12:30: QBS

School of Physical Sciences - Wed, 29/01/2025 - 16:44
QBS

Abstract not available

Add to your calendar or Include in your list

Mon 28 Apr 12:30: QBS

School of Physical Sciences - Wed, 29/01/2025 - 16:43
QBS

Abstract not available

Add to your calendar or Include in your list