Events: Matematiska vetenskaperhttp://www.chalmers.se/sv/om-chalmers/kalendariumUpcoming events at Chalmers University of TechnologyMon, 05 Dec 2022 11:06:55 +0100http://www.chalmers.se/sv/om-chalmers/kalendariumhttps://www.chalmers.se/en/departments/math/calendar/Pages/Examensarbete221206.aspxhttps://www.chalmers.se/en/departments/math/calendar/Pages/Examensarbete221206.aspxMaster Thesis presentation<p>MV:L14 och Zoom</p><p>Sævar Óli Valdimarsson: Prediction of mass transport properties in 3D microstructures using 2D CNNs</p>https://www.chalmers.se/en/departments/math/calendar/Pages/KASS-seminar221206.aspxhttps://www.chalmers.se/en/departments/math/calendar/Pages/KASS-seminar221206.aspxKASS seminar<p>MV:L15, Chalmers tvärgata 3</p><p>Mingchen Xia: Intersection theory on Riemann—Zariski spaces and Chern—Weil formulae</p><br />Abstract: Given a holomorphic Hermitian vector bundle (E,h) on a smooth complex variety X, the classical Chern—Weil formula says that the Chern forms of (E,h) represent the Chern classes of E. When h has singularities, the corresponding result fails. When h is positively curved, there are many different ways to make sense of the Chern forms/currents of (E,h). For our problem, the most natural one is the non-pluripolar theory. I will explain the Chern—Weil formulae of the following form: the non-pluripolar Chern currents of singular Hermitian vector bundles represent some algebraic intersection numbers. As we will see, this can not be done using the intersection theory on X. Instead, we need to develop an appropriate intersection theory on the Riemann—Zariski space and interpret the algebraic intersection theory properly. https://www.chalmers.se/en/centres/chair/events/AI-Ethics/Pages/AI-Ethics-with-Emma-Engstrom.aspxhttps://www.chalmers.se/en/centres/chair/events/AI-Ethics/Pages/AI-Ethics-with-Emma-Engstrom.aspxAI Ethics with Emma Engström<p>Online, register to receive the link</p><p>A new world with AI-driven decisions – implications for privacy, autonomy, and democracy</p><div><img src="/SiteCollectionImages/20220701-20221231/Emma%20Engström.jpg" alt="Photo of Emma Engström" class="chalmersPosition-FloatRight" style="margin:5px" /><br />AI has recently transformed from an obscure academic endeavor to a driver of the fourth industrial revolution. With ever-advancing machine learning algorithms, AI can make ever-more precise inference using an ever-expanding set of behavioral data online. In particular, AI that functions as decision-support has developed and spread amazingly fast.</div>
<div><br /></div>
<div>This presentation articulates some of the ensuing social challenges. It accounts for work in progress within the project <em>Predicting the diffusion of artificial intelligence (WASP-HS)</em>, in which we ask:</div>
<div>How are AI-technologies adopted in society? Do they diffuse similarly as earlier general-purpose technologies? Do concerns about privacy, autonomy, and fairness influence potential adopters of AI? What are the social and ethical implications of the AIs that are likely to spread particularly fast?</div>
<br /><div><strong>Emma Engström</strong> is a researcher within the theme <em>The societal impact of new technologies</em> at the Institute for Futures studies in Stockholm, and a researcher at the Department of Urban planning and Environment (KTH). She holds a PhD in Environmental Engineering (KTH), an MSc in Engineering Physics (KTH), and a BSc in Political Science (Uppsala University). Her research interests span broadly across topics within Technology in Society.</div>
<div><span id="ms-rterangecursor-start"><br /></span></div>
<div><a href="https://ui.ungpd.com/Surveys/2dfca98a-c795-4a7f-b25d-559c65919c06" target="_blank"><img class="ms-asset-icon ms-rtePosition-4" src="/_layouts/images/icgen.gif" alt="" />Register here to receive the link to the seminar</a><br /></div>https://www.chalmers.se/en/departments/math/calendar/Pages/TLM-seminarium221207.aspxhttps://www.chalmers.se/en/departments/math/calendar/Pages/TLM-seminarium221207.aspxTLM seminar<p>MV:L15 and Zoom</p><p>Collaboration seminar: Exchange of ideas and collaboration between teacher groups in Mathematics and Physics</p><br />The seminar is held in Swedish.<div>Vi samlar personer på Matematiska vetenskaper och på Institutionen för fysik som är intresserade av utveckling av och forskning kring lärande. Målet med seminariet är att börja dela erfarenheter och lära av varandra samt finna former för att fortsätta samverka. </div>https://www.chalmers.se/en/departments/math/calendar/Pages/Computational221207.aspxhttps://www.chalmers.se/en/departments/math/calendar/Pages/Computational221207.aspxComputational and Applied Mathematics (CAM) seminar<p>MV:L14 and Zoom</p><p>Jan S Hesthaven, EPFL: Digital Twins through Reduced Order Models and Machine Learning</p><br />Abstract:
The vision of building digital twins for complex infrastructure and systems is old. However, realizing it remains very challenging due to the need to combine advanced computational modeling, reduced order models, data infusion for calibration, updating and uncertainty management, and sensor integration to obtain models with true predictive value for decision support. Nevertheless, the perspectives of using digital twins for predictive maintenance, operational optimization, and risk analysis are very substantial and the potential for impact significant, from safety, planning, and financial points of view. <div>In this talk we shall first discuss the importance of reduced models in the development of digital twin technologies and continue by discussing different aspects of the challenges associated with developing digital twins through a few examples, combining advanced model and data driven technologies, e.g., classifiers, Gaussian regression and neural networks, to enable failure analysis, optimal sensor placement and, time permitting, multi-fidelity methods and risk analysis for rare events. </div>
<div>These are all elements of the workflow that needs to be realized to address the challenge of building predictive digital twins and we shall demonstrated the value of such technologies through a number of different examples of increasing complexity.</div>https://www.chalmers.se/en/departments/math/calendar/Pages/Provforelasningar221208.aspxhttps://www.chalmers.se/en/departments/math/calendar/Pages/Provforelasningar221208.aspxTrial lectures, senior lectureship in algebra and geometry<p>Pascal, Mallvinden (zoom)</p><p>Lars Kühne, Lars Martin Sektnan, Shuntaro Yamagishi</p><br />9.00 Lars Kühne (Pascal)<p></p>
<p>10.00 Lars Martin Sektnan (Pascal)</p>
<p> 14.30 Shuntaro Yamagishi (zoom) <a href="https://gu-se.zoom.us/j/65729952293">https://gu-se.zoom.us/j/65729952293</a></p>
<p>The lecture in itself is about 20 minutes long and begins 10 minutes after the given time. Then, it is 10 minutes time for questions. The subject is "An introduction to Green's and Stokes' theorems".</p>
https://www.chalmers.se/en/departments/math/calendar/Pages/Statistics221208.aspxhttps://www.chalmers.se/en/departments/math/calendar/Pages/Statistics221208.aspxStatistics seminar<p>MV:L14, Chalmers tvärgata 3</p><p>Karin Hårding and Daire Carroll, University of Gothenburg: Population dynamics and ecology of seal populations, empirical data and the search for theory to help our understanding. Stochastic growth models, image analysis, spatial distribution and telemetry data on migrations</p><br />Abstract: This talk is about how statistical and mathematical methods are helpful when we try to understand processes in wildlife populations. The European harbour seal (Sw: knubbsälen) has been studied carefully for 40 years and the long time series allows analysis of how population growth is regulated. Recently the population growth has declined and we visited the colonies to try to document in detail what is going on in order to give better advise to managers. We develop new methods for estimating body size from drones and for counting seals from photos with machine learning algorithms. We apply stochastic population growth models, dynamic energy budget models, and we discuss what is density dependence in age structured populations in a variable environment. We are also interested in new collaborations and feed back and look forward to interesting discussions on ways forward. Welcome! Karin and Dairehttps://www.chalmers.se/en/areas-of-advance/ict/calendar/Pages/AI4Science-with-Tess-E.-Smidt.aspxhttps://www.chalmers.se/en/areas-of-advance/ict/calendar/Pages/AI4Science-with-Tess-E.-Smidt.aspxAI4Science with Tess E. Smidt<p>Online</p><p>Euclidean Symmetry Equivariant Machine Learning – Overview, Applications, and Open Questions</p><img src="/SiteCollectionImages/Centrum/CHAIR/events/Tess%20Smidt%20webb.jpg" class="chalmersPosition-FloatRight" alt="Photo of Tess Smidt" style="margin:5px" /><br />Atomic systems (molecules, crystals, proteins, etc.) are naturally represented by a set of coordinates in 3D space labeled by atom type. This is a challenging representation to use for machine learning because the coordinates are sensitive to 3D rotations, translations, and inversions (the symmetries of 3D Euclidean space).<br /><br /><div>In this talk I’ll give an overview of Euclidean invariance and equivariance in machine learning for atomic systems. Then, I’ll share some recent applications of these methods on a variety of atomistic modeling tasks (ab initio molecular dynamics, prediction of crystal properties, and scaling of electron density predictions). Finally, I’ll explore open questions in expressivity, data-efficiency, and trainability of methods leveraging invariance and equivariance.</div>
<br /><strong>Tess Smidt</strong> is an Assistant Professor of Electrical Engineering and Computer Science at MIT. Tess earned her SB in Physics from MIT in 2012 and her PhD in Physics from the University of California, Berkeley in 2018. Her research focuses on machine learning that incorporates physical and geometric constraints, with applications to materials design.<br /><br />Prior to joining the MIT EECS faculty, she was the 2018 Alvarez Postdoctoral Fellow in Computing Sciences at Lawrence Berkeley National Laboratory and a Software Engineering Intern on the Google Accelerated Sciences team where she developed Euclidean symmetry equivariant neural networks which naturally handle 3D geometry and geometric tensor data.<br /><br /><a href="https://chalmers.zoom.us/j/67663462974" target="_blank"><img class="ms-asset-icon ms-rtePosition-4" src="/_layouts/images/icgen.gif" alt="" />Connect via Zoom</a><br /><strong>Password:</strong> ai4sciencehttps://www.chalmers.se/en/centres/chair/events/AI-Ethics/Pages/AI-Ethics-with-Gabriel-Skantze.aspxhttps://www.chalmers.se/en/centres/chair/events/AI-Ethics/Pages/AI-Ethics-with-Gabriel-Skantze.aspxAI Ethics with Gabriel Skantze<p>Online, register to receive the link</p><p>Human-likeness in robotics and conversational AI: Opportunities and risks</p><div><img src="/SiteCollectionImages/20220701-20221231/Gabriel%20Skantze.jfif" alt="Photo of Gabriel Skantze" class="chalmersPosition-FloatRight" style="margin:5px" /><br />Many researchers are working on how to make robots and conversational systems look and sound like humans. Other than the intriguing challenges involved, there are many potential advantages in doing so, one of them being that we already know how to interact with each other, so communicating with human-like systems should be intuitive for us. However, there are also potential disadvantages and risks involved. In this talk, I will present our own research on these aspects and discuss how we can approach the question in a more systematic way. </div>
<div><br /></div>
<div><strong>Gabriel Skantze</strong> is a Professor in Speech Technology at KTH Royal Institute of Technology. He is leading several research projects on Conversational AI and Human-Robot Interaction. He is also co-founder and chief scientist of the social robotics company Furhat Robotics.</div>
<div><br /></div>
<div><a href="https://ui.ungpd.com/Surveys/6644e3ff-a3e0-40df-aa50-106bc58cadbf" target="_blank"><img class="ms-asset-icon ms-rtePosition-4" src="/_layouts/images/icgen.gif" alt="" />Register here to receive the link to the seminar</a><br /></div>https://www.chalmers.se/en/departments/math/calendar/Pages/Statistics221215.aspxhttps://www.chalmers.se/en/departments/math/calendar/Pages/Statistics221215.aspxStatistics seminar<p>MV:L14, Chalmers tvärgata 3</p><p>Johan Jonasson, Chalmers University and University of Gothenburg: Noise sensitivity/stability for deep Boolean neural nets</p><br />Abstract: A well-known and ubiquitous property of neural net classifiers is that they can be fooled into misclassifying some objects by changing the input in tiny ways that are indistinguishable for the human eye. These changes can be adversarial, but sometimes they can be just random noise. This makes it interesting to ask if this property is something that almost all neural nets have and, when they do, why that is. There are good heuristic explanations, but to prove mathematically rigorous results seems very difficult in general. Here we prove some first results on various toy models. We treat our questions within the framework of the established field of noise sensitivity/stability. What we prove can roughly be stated as: <div><ul><li>A sufficiently deep fully connected network with sufficiently wide layers and iid Gaussian weights is noise sensitive, i.e. an arbitrarily small random noise makes the predicted class of a binary input string before and after the noise is added virtually independent. If one imposes correlations on the weights corresponding to the same input features, this still holds unless the correlation is very close to 1. </li>
<li>Neural nets consisting of only convolutional layers may or may not be noise sensitive and we present examples of both behaviours.</li></ul>
</div>https://www.chalmers.se/en/departments/math/calendar/Pages/Licentiatseminarium221220.aspxhttps://www.chalmers.se/en/departments/math/calendar/Pages/Licentiatseminarium221220.aspxLicentiate seminar in bioscience<p>Pascal, Hörsalsvägen 1</p><p>David Lund: Computational discovery of antibiotic resistance genes and their horizonal transfer</p><br />Introducer: Professor Thomas Nordahl Petersen, Danmarks Tekniske Universitet<br />https://www.chalmers.se/en/departments/math/calendar/Pages/Disputation230119.aspxhttps://www.chalmers.se/en/departments/math/calendar/Pages/Disputation230119.aspxThesis defence in mathematics<p>Pascal, Hörsalsvägen 1</p><p>Stepan Maximov: Infinite-dimensional Lie bialgebras and Manin pairs</p><br />Faculty opponent: Professor Volodymyr Mazorchuk, Uppsala Universityhttps://www.chalmers.se/en/departments/math/calendar/Pages/Disputation230120.aspxhttps://www.chalmers.se/en/departments/math/calendar/Pages/Disputation230120.aspxThesis defence in mathematics<p>Pascal, Hörsalsvägen 1</p><p>Kristian Holm: Limit Theorems for Lattices and L-functions</p><br />Faculty opponent: Professor Florent Jouve, Université de Bordeaux, Francehttps://www.chalmers.se/en/departments/math/calendar/Pages/nordstat-2023.aspxhttps://www.chalmers.se/en/departments/math/calendar/Pages/nordstat-2023.aspx29th Nordic Conference in Mathematical Statistics<p>Kemihuset, Kemigården 4</p><p>(NORDSTAT 2023)</p><br />NORDSTAT will be an in-person conference consisting of plenary lectures, invited and contributed talks, and a poster session. A preliminary program is available.