Pre-recorded Lectures

Over the next few weeks, we will be releasing lectures on various machine learning subjects on the Fields youtube channel. Please click the subjects below to enjoy the lectures.

Introduction

Dynamical Systems for Machine Learning

• Anas Barakat - Convergence and Dynamical Behavior of the ADAM Algorithm for Non Convex Stochastic Optimization
• Weinan E - Machine Learning via Dynamical Systems
• Maximilian Engel - Optimization driven dynamics with stochastic noise in games
• Fryderyk Falniowski - Robust routes to chaos for online learning in congestion games
• Guy Gur-Ari - The catapult phase of deep learning
• Liam Hodgkinson - Multiplicative noise and heavy tails in stochastic optimization
• Soon Hoe Lim - Understanding Recurrent Neutral Networks Using Response Theory
• Wei Kang - Data Generation for Deep Learning in Model-Based Optimal Feedback Design
• Anna Korba - Maximum Mean Discrepancy Gradient Flow
• Alexis Laignelet - Towards Robust and Stable Deep Learning Algorithms for Forward-Backward Stochastic Differential Equations
• Qianxiao Li - On the Curse of Memory in Linear Recurrent Neural Networks
• Zhenyu Liao - Dynamical aspects of learning linear neural networks
• Michael Mahoney - Continuous-in-Depth Neural Networks
• W. Garrett Mitchener - Ranking with Hamiltonian dynamics
• Georgios Piliouras - Learning in Zero-Sum Games: Continuous vs Discrete Time Dynamics
• Viktor Reshniak - Robust learning with implicit residual networks
• Adil Salim - Primal Dual interpretation of the Proximal Gradient Langevin algorithum
• Alex Sherstinsky - Recurrent Neural Networks From Differential Equations And Signal Processing Perspectives
• Kelly Spendlove - Conley theory and the global dynamics of games
• Molei Tao - Stochasticity of Deterministic Gradient Descent: Large Learning Rate for Multiscale Objective Function
• Haijun Yu - OnsagerNet: Learning Stable and Interpretable Dynamics using a Generalized Onsager Principle
• Benjamin J. Zhang - Sampling via Controlled Stochastic Dynamical Systems
• Enrique Zuazua and Borjan Geshkovski - Turnpike Control and Deep Learning

Data-driven Modelling

• Philip J. Aston - Transforming Signals to Images Using Attractor Reconstruction for Deep Learning
• J.-F. Barczi and Emilie Peynaud - Data-driven PDE modelling: Trick or treat!
• Andreas Bittracher - Probabilistic aggregation of large under-sampled Markov chains
• Steve Brunton - Interpretable and Generalizable Machine Learning for Modeling and Control
• Karim Cherifi - Data Driven Port Hamiltonian systems modelling and control
• Daan Crommelin - Resampling with neural networks for data-driven stochastic parameterization
• Bachir El Khadir - Learning Dynamical Systems with Side Information
• Benjamin Herrmann - Modeling synchronization in forced turbulent oscillator flows
• Oliver Junge - Linear response for the dynamic Laplacian and finite-time coherent sets
• Kadierdan Kaheman - SINDy-PI: A Robust Algorithym for Parallel Implicit Sparse Identification of Nonlinear Dynamics
• Stefan Klus - Data-driven approximation of the Koopman generator and Shrodinger operator
• Matthew Levine - Machine-learning of model error in ODEs
• Mohammad Farazmand - Data-Driven Prediction of Multistable Systems from Sparse Measurements
• Andrew Fraser - Hidden Markov Models and Dynamical Systems
• Ty Frazier - Challenges for Building Surrogate Model for Nuclear Reaction Systems
• Gary Froyland - Optimising linear response of kernel dynamical systems and operator-theoretic extraction of the ENSO cycle
• Georg Gottwald - Supervised learning from noisy observations: Combining machine-learning techniques with data assimilation
• John Harlim - Learning Missing Dynamics from Data
• Roman Khotyachuk - Big Data and Machine Learning for Analysis of Numerical PDEs
• Péter Koltai - Collective variables in complex systems
• J. Nathan Kutz - Data-driven learning of control signals, parameters, and governing equations
• Martin Lellep - Interpreted machine learning in fluid dynamics: Explaining relaminarisation events in wall-bounded shear flows
• Matthew Levine - Machine-learning of model error in ODEs
• Kevin Lin - Data driven model reduction and the Koopman-Mori-Zwanzig formalism
• Romit Maulik - Machine learning enablers for system optimization and design
• Arash Mehrjou - Learning Dynamical Systems using Local Stability Priors
• Rebecca Morrison - Learning sparse dynamics of interacting systems
• Frank Noé - Variational methods and deep learning for high-dimensional dynamical systems
• Feliks Nüske - Gedmd: Data-Driven Analysis of Stochastic Dynamical Systems
• Vlachas Pantelis - Recurrent Neural Networks for Spatiotemporal Prediction of Chaotic Dynamics
• Themis Sapsis - Output-Weighted Active Sampling for Uncertainty Quantification and Prediction of Rare Events
• Gowtham S. Seenivasaharagavan - On mean subtraction and Dynamic Mode Decomposition
• Felix X.-F Ye - Nonparametric Nonlinear Model Reduction for slow-fast SDEs near manifolds

Deep Learning

• Anastasia Borovykh - Neural Networks for Solving PDEs
• Marco Gallieri - Safe AI with control theory
• Lars Gruene - Computing Lyapunov functions via neural networks avoiding the curse of dimensionality
• Guillaume Lajoie - Recurrence vs Attention: untangling tradeoffs in self-attentive neural networks
• Ofir Lindenbaum - Local Conformal AutoEncoder
• Houman Owhadi - Do ideas have shape? Plato's theory of forms as the continuous limit of artificial neutral networks
• Zhi-Qin John Xu - Frequency Principle in Deep Learning

Kernel Methods

• Yifan Chen - Consistency of Hierarchical Parameter Learning Emperical Bayes and Kernel Flow Approaches
• Matthieu Darcy - Kernel Flows Demystified: Application to Regression
• Stefano De Marchi - Variably Scaled Discontinuous Kernels (VSDK): basics and some applications (3 talks)
• Peter Giesl - Approximation of matrix-valued functions with applications to Contraction Metrics
• Krithika Manohar - Kernel Analog Forecasting for Multiscale Problems
• Nicholas H. Nelsen - The Random Feature Model for Input-Output Maps Between Function Spaces
• Hung Nguyen - Gaussian Process Learning for Power Systems
• Juan Orduz - Gaussian Processes for Time Series Forecasting
• Amir Sagiv - Learning Model Parameters with an Unknown Observation Function
• Florian Schäfer - Sparse Cholesky factorization by Kullback-Leibler minimization
• George Wynne - A Kernel Two-Sample Test For Functional Data
• Gene Ryan Yoo - Learning patterns with kernels and learning kernels from patterns

Learning Theory

• Stefano Luzzatto - Rigorous numerics for critical orbits in the quadratic family
• Matt Thorpe - On the Number of Labels in Semi-Supervised Learning

Transfer and Koopman operators

• Marvyn Gulina - Two methods to approximate the Koopman operator with a reservoir computer

Reservoir Computing

• Erik Bollt - On Explaining the Surprising Success of Reservoir Computing Forecaster of Chaos?
• Thomas L. Carroll - Network Statistics for Reservoir Computing
• Andrea Ceni - Multistability in input-driven recurrent neural networks
• Davide Faranda - Boosting performance in Machine Learning of Turbulent and Geophysical Flows via scale separation
• Daniel J. Gauthier - Reservoir Computing with Autonomous Boolean Networks on Field Programmable Gate Arrays
• Allen G. Hart - Embedding and Approximation Theorems for Echo State Networks
• Daniel P. Lathrop - Reservoir computing: prediction and high-speed hardware accelerators
• Juan-Pablo Ortega - Explaining the reservoir computing phenomenon using randomized discrete-time signatures
• Graham Rowlands - Reservoir Computing with Superconducting Circuits
  
• Josef Teichmann - Randomized Signature and Reservoir Computing with application to Finance
• Pietro Verzelli - Learn to Synchronize, Synchronize to Learn: measuring the Echo State Property
• Lucas Zipp - Time- and Wavelength-Multiplexed Photonic Reservoir Computing

Geometric and Topological Data Analysis

• Paweł Dłotko - Geometry and topology of data
• Niklas Hellmer - New metrics for persistence diagrams
• Woojin Kim - Spatiotemporal Persistent Homology for Dynamic Metric Spaces
• Arthur J. Krener - Manifold Reconstruction by Simplicial Nonlinear Principal Component Analysis (SNPCA)
• Mateusz Przybylski - Towards the Conley Index for Binary Relations on Data
• Sarah Tymochko - Using Zigzag Persistence for Bifurction Analysis

Signature-based methods and related topics

• Vidit Nanda - Signature Features in Topological Data Analysis

Miscellaneous Techniques

• Raphael Gerlach - Set oriented path following of dependent attractors