**Seminars**

**May 10, 2023: 16:15, Research Seminar Data Science Foundations**

Giulia Bertagnolli (University of Trento)

Random walks on networks (toward information geometry)

Location: Blohmstraße 15 (HIP One), 5th Floor, Room 5.002

Abstract

Complex physical and social systems find a handy representation in terms of graphs, which, in this context, are called complex networks. Entities in these systems naturally “communicate”, or exchange “information”, e.g., a group of people interacting via email or sharing links, liking posts, and following each other on social platforms, exchange information as part of their social life. Neurons, connected by synapses and fibre bundles, exchange of neuro-physiological signals, enabling cognition. In fish schools, aggregations of fish, who come together in an interactive, social way, the (possibly passive) communication between fish allows them to act as a super-system. All complex systems show some emergent behaviour that cannot be ascribed to the actions and behaviour of their individual components. This emergent behaviour is a function of both the interaction patterns, i.e. the links in the graph, and the communication strategy, which can be modelled as a dynamical process on the network. In this talk, we will see, firstly, how Markovian random walks on networks model diffusion dynamics in the complex system and why this approach is useful in network science. Then, we will see an example of non-Markovian random walk, which mimics the run-and-tumble motion of bacteria. Eventually, it should become clear how this led me here, trying to learn information geometry.

**April 19, 2023: 15:00, Research Seminar Data Science Foundations**

Jabob J. W. Bakermans (University of Oxford)

Compositional planning by making memories of the future

Location: Blohmstraße 15 (HIP One), 5th Floor, Room 5.002

Abstract

Hippocampus is critical for memory, imagination, and constructive reasoning. However, recent models have suggested that its neuronal responses can be well explained by state-spaces that model the transitions between experiences. How do we reconcile these two views? I’ll show that if state-spaces are constructed compositionally from existing primitives, hippocampal responses can be interpreted as compositional memories, binding these primitives together. Critically, this enables agents to behave optimally in novel environments with no new learning, inferring behaviour directly from the composition. This provides natural interpretations of generalisation and latent learning. Hippocampal replay can build and consolidate these compositional memories, but importantly, due to their compositional nature, it can construct states it has never experienced – effectively building memories of the future. This enables new predictions of optimal replays for novel environments, or after structural changes.

**March 17, 2023: 15:00, Research Seminar Data Science Foundations**

Minh Ha Quang (RIKEN Center for Advanced Intelligence Project (AIP))

An information geometric and optimal transport framework for Gaussian processes

Location: Blohmstraße 15 (HIP One), 5th Floor, Room 5.002

Abstract

Information geometry (IG) and Optimal transport (OT) have been attracting much research attention in various fields, in particular machine learning and statistics. In this talk, we present results on the generalization of IG and OT distances for finite-dimensional Gaussian measures to the setting of infinite-dimensional Gaussian measures and Gaussian processes. Our focus is on the Entropic Regularization of the 2-Wasserstein distance and the generalization of the Fisher-Rao distance and related quantities. In both settings, regularization leads to many desirable theoretical properties, including in particular dimension-independent convergence and sample complexity. The mathematical formulation involves the interplay of IG and OT with Gaussian processes and the methodology of reproducing kernel Hilbert spaces (RKHS). All of the presented formulations admit closed form expressions that can be efficiently computed and applied practically. The theoretical formulations will be illustrated with numerical experiments on Gaussian processes.

**February 01, 2023: 15:00, Research Seminar Data Science Foundations**

Christian Gumbsch (Max Planck Institute for Intelligent Systems and University of Tübingen)

Events – Learning Latent Codes for Hierarchical Prediction and Generalization

Location: Blohmstraße 15 (HIP One), 5th Floor, Room 5.002

**19.10.2022: 15:00 Uhr,** **Research Seminar Data Science Foundations**

Nihat Ay (Hamburg University of Technology)

Die Klugheit der Dinge

Location: Blohmstraße 15 (HIP One), 5th Floor, Room 5.002

More info: Stud.IP

**17.11.2021: 9:30 Uhr,** **Research Seminar Data Science Foundations**

Johannes Rauh (MPI for Mathematics in the Sciences, Leipzig and Federal Institute for Quality and Transparency in Healthcare, Berlin)

Uncertainty and Stochasticity of Optimal Policies

Location: Blohmstraße 15 (HIP One), 5th Floor, Room 5.002

Abstract

We are interested in optimal action selection mechanisms, policies, that maximize an expected long term reward. Our main model are POMDPs (Partially Observed Markov Decision Problems). While the optimal policy can be stochastic in the general case, we find conditions under which the optimal policy is deterministic, at least for some observations, or under which the stochasticity can be bounded. This talk presents joint work with Guido Montúfar and Nihat Ay.

**14.06.2021, 17:00 Uhr, Seminar within the Machine Learning in Engineering initiative MLE@TUHH**

Nihat Ay (Hamburg University of Technology)

Information Geometry for Deep Learning

**The following seminars were co-organised and took place at the **Max Planck Institute for Mathematics in the Sciences.

**19.03.2021, 16:00 Uhr, Mathematics of Data Seminar**

Stefanie Jegelka *(Machine Learning Group at MIT, USA)*

Representation and Learning in Graph Neural Networks

The seminar is cancelled.

**04.08.2020, 11:00 Uhr, Mathematics of Data Seminar**

Marco Mondelli *(IST Austria)*

Understanding Gradient Descent for Over-parameterized Deep Neural Networks

**20.07.2020, 17:00 Uhr, Mathematics of Data Seminar**

Franca Hoffmann *(California Institute of Technology)*

Kalman-Wasserstein Gradient Flows

**04.02.2020, 11:00 Uhr, Special Seminar**

Xerxes Arsiwalla *(Pompeu Fabra University Barcelona, Spain)*

Extending Integrated Information Theories for Cognitive Systems

**10.12.2019, 16:45 Uhr, Chalk Talk – Mathematics of Data Seminar**

Mikhail Belkin *(The Ohio State University, USA)*

Chalk Talk: What’s next for machine learning? Some thoughts toward a unified theory of supervised inference.

**14.11.2019, 11:00 Uhr, Mathematics of Data Seminar**

Kathlén Kohn *(KTH Royal Institute of Technology, Stockholm)*

The geometry of neural networks

**23.10.2019, 11:00 Uhr, Mathematics of Data Seminar**

Věra Kůrková *(Institute of Computer Science, Czech Academy of Sciences, Czech Republic)*

Lower Bounds on Complexity of Shallow Networks

**18.09.2019, 11:00 Uhr, Mathematics of Data Seminar**

Vladimir Temlyakov *(University of South Carolina)*

Supervised learning and sampling error of integral norms in function classes

**16.07.2019, 11:00 Uhr, Mathematics of Data Seminar**

Lamiae Azizi *(The University of Sydney)*

A Mathematical trip into the Data Science realm

This seminar is cancelled.

**28.05.2019, 11:15 Uhr, Mathematics of Data Seminar**

Nicolas Garcia Trillos *(Department of Statistics, University of Wisconsin-Madison, USA)*

The use of geometry to learn from data, and the learning of geometry from data.

**10.04.2019, 11:00 Uhr, Mathematics of Data Seminar**

Gabriel Peyré *(CNRS and Ecole Normale Supérieure, Paris, France)*

Computational Optimal Transport for Data Sciences

**07.03.2019, 11:00 Uhr, Mathematics of Data Seminar**

Stefania Petra *(Universität Heidelberg)*

Compressed Sensing – From Theory To Practice

**14.02.2019, 11:00 Uhr, Mathematics of Data Seminar**

Felix Krahmer *(Technische Universität München)*

Blind deconvolution with randomness – convex geometry and algorithmic approaches

**28.01.2019, 11:00 Uhr, Mathematics of Data Seminar**

Nils Bertschinger *(Frankfurt Institute for Advanced Studies – FIAS, Germany)*

A geometric structure underlying stock correlations

**08.11.2018, 11:00 Uhr, Mathematics of Data Seminar**

Benjamin Fehrmann *(University of Oxford)*

Convergence rates for mean field stochastic gradient descent algorithms

**27.09.2018, 11:00 Uhr, Mathematics of Data Seminar**

Max von Renesse *(Universität Leipzig)*

Topics in Deterministic and Stochastic Dynamical Systems on Wasserstein Space

**14.08.2018, 16:30 Uhr, Mathematics of Data Seminar**

Afonso Bandeira *(Courant Institute of Mathematical Sciences, New York)*

Statistical estimation under group actions: The Sample Complexity of Multi-Reference Alignment

**11.07.2018, 15:30 Uhr, Mathematics of Data Seminar**

Harald Oberhauser *(University of Oxford)*

Learning laws of stochastic processes

**18.06.2018, 15:30 Uhr, Mathematics of Data Seminar**

Anna Seigal *(University of California, Berkeley)*

Structured Tensors and the Geometry of Data

**14.05.2018, 14:00 Uhr, Seminar on Theory of Embodied Intelligence**

Keyan Ghazi-Zahedi *(MPI MIS, Leipzig)*

Quantifying Morphological Computation

**02.05.2018, 11:00 Uhr, Mathematics of Data Seminar**

Steffen Lauritzen *(University of Copenhagen, Denmark)*

Max-linear Bayesian networks

**24.04.2018, 15:30 Uhr, Mathematics of Data Seminar**

Benjamin Recht *(University of California, Berkeley)*

The statistical foundations of learning to control

**22.03.2018, 14:00 Uhr, Information Geometry Seminar**

Dimitri Marinelli *(Romanian Institute of Science and Technology – RIST, Romania)*

Quantum Information Geometry and Boltzmann Machines

**08.01.2018, 14:00 Uhr, LikBez Seminar**

Nihat Ay *(MPI MIS, Leipzig)*

Causal Inference II

**04.12.2017, 11:00 Uhr, Special Seminar**

Wolfgang Löhr *(TU Chemnitz)*

Continuum limits of tree-valued Markov chains and algebraic measure trees

**27.11.2017, 14:00 Uhr, Seminar on Theory of Embodied Intelligence**

Fabio Bonsignorio *(Scuola Superiore Sant’Anna, Pisa, Italy)*

Modeling of Networked Embodied Cognitive Processes

**10.11.2017, 11:45 Uhr, Information Geometry Seminar**

Jun Zhang *(University of Michigan-Ann Arbor, USA)*

Statistical Manifold and Entropy-Based Inference

**16.10.2017, 14:00 Uhr, Information Geometry Seminar**

Luigi Malagò *(Romanian Institute of Science and Technology – RIST, Romania)*

From Natural Gradient to Riemannian Hessian: Second-order Optimization over Statistical Manifolds

**10.05.2017, 14:00 Uhr, Information Geometry Seminar**

Domenico Felice *(University of Camerino, Italy)*

Hamilton-Jacobi approach to Potential Functions in Information Geometry

**25.11.2016, 15:30 Uhr, Special Seminar**

František Matúš *(Czech Academy of Sciences, Prague, Czech Republic)*

Polyquantoids and quantoids: quantum counteparts of polymatroids and matroids

**28.06.2016, 15:30 Uhr, Seminar on Theory of Embodied Intelligence**

Daniel Polani *(University of Hertfortshire, United Kingdom)*

On Information and the Drivers of Cognition

**10.05.2016, 11:00 Uhr, Seminar on Theory of Embodied Intelligence**

Roy Fox *(School of Computer Science and Engineering, Hebrew University, Israel)*

Minimum-Information Planning in Partially-Observable Decision Problems

**30.03.2016, 11:00 Uhr, Seminar on Theory of Embodied Intelligence**

Daniel Häufle *(Stuttgart Research Center for Simulation Technology, University of Stuttgart, Germany)*

Musculo-Skeletal Models of Human Movement: Tools to Quantify Embodiment

**21.01.2016, 11:00 Uhr, Seminar on Theory of Embodied Intelligence**

Daniel Häufle *(Stuttgart Research Center for Simulation Technology, University of Stuttgart, Germany)*

Musculo-Skeletal Models of Human Movement: Tools to Quantify Embodiment

This talk is canceled!

**09.11.2015, 14:00 Uhr, Arbeitsgemeinschaft NEURONALE NETZE UND KOGNITIVE SYSTEME**

Tomas Veloz *(University of British Columbia, Canada)*

Toward a Quantum Theory of Cognition: History, Development and Perspectives

**08.04.2015, 14:00 Uhr, Special Seminar**

Sajad Saeedinaeeni *(Universität Leipzig)*

On asymptotic optimality of ML-type detectors in quantum hypothesis testing

**17.03.2015, 14:00 Uhr, Special Seminar**

Peter Gmeiner *(Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany)*

Information-Theoretic Cheeger Inequalities

**10.03.2015, 11:00 Uhr, Seminar on Theory of Embodied Intelligence**

Benjamin Friedrich *(Max-Planck-Institut für Physik komplexer Systeme, Dresden)*

Intelligent motility control of biological swimmers

**18.02.2015, 14:00 Uhr, Special Seminar**

Ryszard Kostecki *(Perimeter Institute for Theoretical Physics, Waterloo, Canada)*

Quantum information geometry as a foundation for quantum theory beyond quantum mechanics

**19.01.2015, 11:00 Uhr, Seminar on Theory of Embodied Intelligence**

Oliver Brock *(Technische Universität Berlin, Robotics and Biology Laboratory)*

Towards an Alchemy of Intelligence

**15.01.2015, 10:30 Uhr, Special Seminar**

František Matúš *(Academy of Sciences of the Czech Republic, Institute of Information Theory and Automation)*

Algebraic Problems Related to Entropy Regions