Hide menu

IDA Machine Learning Seminars - Spring 2023

Wednesday, February 22, 3.15 pm, 2023

How Epidemiology and Behavioural Trials are (slowly) adopting the Bayesian Paradigm and Causal Inference Language
Marcus Bendtsen, Department of Health, Medicine and Caring Sciences (HMV), Linköping University

Abstract: In this seminar, Dr. Marcus Bendtsen will give a progress report on his mission to introduce (force) the Bayesian paradigm into public health research (and widely in medical research). He will give real-world examples of trials that would have benefitted from Bayesian analysis rather than the prevailing null-hypothesis testing framework so often adopted in medicine and public health, and show examples of more recent trials which have been designed from the bottom up to be "Bayesian". Dr. Bendtsen will also discuss how causal graphs are powerful for teaching basic concepts and challenges of epidemiological research to those who do not have any prior experience, and how even simple causal graphs (and inference) can have a lot of value in public health.
Location: Ada Lovelace and also on Zoom
Organizers: Fredrik Lindsten, Sourabh Balgi

Wednesday, March 22, 3.15 pm, 2023

Text Classification with Born's Rule
Emanuele Guidotti, Institute of Financial Analysis, University of Neuchâtel

Abstract: This paper presents a text classification algorithm inspired by the notion of superposition of states in quantum physics. By regarding text as a superposition of words, we derive the wave function of a document and we compute the transition probability of the document to a target class according to Born's rule. Two complementary implementations are presented. In the first one, wave functions are calculated explicitly. The second implementation embeds the classifier in a neural network architecture. Through analysis of three benchmark datasets, we illustrate several aspects of the proposed method, such as classification performance, explainability, and computational efficiency. These ideas are also applicable to non-textual data.
Location: BL32 Nobel
Organizers: Fredrik Lindsten, Sourabh Balgi

Wednesday, April 19, 3.15 pm, 2023

Title: AI4Science at Microsoft Research
Rianne van den Berg, Principal Researcher, Microsoft Research Amsterdam

Abstract: In July 2022 Microsoft announced a new global team in Microsoft Research, spanning the UK, China and the Netherlands, to focus on AI for science. In September 2022 we announced that we have also opened a new lab in Berlin, Germany, and recently another team in Redmond (USA) joined our initiative. In this talk I will first discuss some of the research areas that we are currently exploring in AI4Science at Microsoft Research, covering topics such as drug discovery, material generation, neural PDE solvers, electronic structure theory. Then I will dive a little deeper into two recent works that were done at AI4Science. First, I will cover work on the use of score-based generative modeling for coarse-graining (CG) molecular dynamics simulations. By training a diffusion model on protein structures from molecular dynamics simulations we show that its score function approximates a force field that can directly be used to simulate CG molecular dynamics. While having a vastly simplified training setup compared to previous work, we demonstrate that our approach leads to improved performance across several small- to medium-sized protein simulations, reproducing the CG equilibrium distribution, and preserving dynamics of all-atom simulations such as protein folding events. Second, I will discuss our recent work on Clifford Neural layers for PDE modeling. The PDEs of many physical processes describe the evolution of scalar and vector fields. In order to take into account the correlation between these different fields and their internal components, we represent these fields as multivectors, which consist of scalar, vector, as well as higher-order components. Their algebraic properties, such as multiplication, addition and other arithmetic operations can be described by Clifford algebras, which we use to design Clifford convolutions and Clifford Fourier transforms. We empirically evaluate the benefit of Clifford neural layers by replacing convolution and Fourier operations in common neural PDE surrogates by their Clifford counterparts on two-dimensional Navier-Stokes and weather modeling tasks, as well as three-dimensional Maxwell equations.
Location: Ada Lovelace
Organizer: Fredrik Lindsten, Sourabh Balgi

Wednesday, May 10, 3.15 pm, 2023

Title: On The Moment Representation of Stochastic Filtering
Zheng Zhao, Division of Systems and Control, Uppsala University

Abstract: Stochastic filters are an important algorithms for estimating the distributions of latent variables conditioned on the observations. In reality, the filtering distributions are hard to compute, hence, we often resort to approximate representations of the distributions that are easy to compute, for example, samples or Gaussian approximations. In this talk, we introduce a representation based on a sequence of moments. We show that this representation is convergent in distribution as we increase the order of moments, and that the moments are easy to compute by using a numerical quadrature method. Furthermore, we numerically show that the performance of the moment filter is comparable to standard particle filters in terms of convergence, parameter estimation, and computation. Please feel free to take a look at our implementation of the filter at github link.
Location: Alan Turing
Organizer: Fredrik Lindsten, Sourabh Balgi

Wednesday, May 17, 3.15 pm, 2023

Title: Scientific Machine Learning - An overview with applications to inverse problems
Ozan Öktem, Associate Professor, The Royal Institute of Technology (KTH), Stockholm

Abstract: Scientific Machine Learning is an emerging research area focused on the opportunities and challenges of machine learning in the context of complex applications across science, engineering, and medicine. Challenges in these fields have attributes that make them very different in nature to computer science applications where data-driven machine learning has found success. The talk will try to provide an overview of Scientific Machine Learning with emphasis on its recent application to solving large-scale ill-posed inverse problems.
Location: Ada Lovelace
Organizers: Fredrik Lindsten, Sourabh Balgi

Page responsible: Fredrik Lindsten
Last updated: 2023-08-24