Selected Invited Talks
"Optimization, Robustness and Attention in Deep Neural Networks: Insights from Random and NTK Feature Models", ISL Colloquium, Stanford, April 2024.
"From Spectral Estimators to Approximate Message Passing... And Back", Algorithmic Structures for Uncoordinated Communications and Statistical Inference in Exceedingly Large Spaces workshop, Banff International Research Station, March 2024.
"Phase Transitions for Spectral Estimators in Generalized Linear Models via Approximate Message Passing", International Zurich Seminar on Information and Communication (IZS), Zurich, March 2024.
"Fundamental Limits of Autoencoders, and Achieving Them with Gradient Methods", Learning and Information Theory (LITH) Workshop, EPFL, March 2024.
"Towards Understanding the Word Sensitivity of Attention Layers: A Study via Random Features", Information Theory and Applications (ITA) Workshop, UCSD, San Diego, February 2024.
"From Spectral Estimators to Approximate Message Passing... And Back", The Mathematics of Data, National University of Singapore (NUS), January 2024.
"Optimization, Robustness and Privacy in Deep Neural Networks: Insights from the Neural Tangent Kernel", Joint TILOS and OPTML++ seminar, MIT, November 2023 (Online).
"From Spectral Estimators to Approximate Message Passing... And Back", Probabilitas seminar, Harvard, November 2023 (Online).
"Three Vignettes on the Mean-Field Analysis of Neural Networks", Workshop on Analytical Approaches for Neural Network Dynamics, Institute Henri Poincaré (IHP), Paris, October 2023.
"Optimization, Robustness and Privacy in Deep Neural Networks: Insights from the Neural Tangent Kernel", Mathematical Information Science Workshop, Lagrange Mathematics and Computing Research Center (LMCRC), Paris, October 2023.
"Mean-field Analysis of Piecewise Linear Solutions for Wide ReLU Networks", International Congress on Industrial and Applied Mathematics (ICIAM), Waseda University, August 2023.
"Precise Asymptotics for Spectral Methods in Generalized Linear Models with Correlated Gaussian Designs", Joint Statistical Meeting (JSM), Toronto, August 2023.
"Inference in High Dimensions", European School of Information Theory (ESIT), University of Bristol, July 2023.
"From Spectral Estimators to Approximate Message Passing... And Back", Workshop on Learning and Inference from Structured Data: Universality, Correlations and Beyond, International Centre for Theoretical Physics (ICTP), July 2023.
"Three Vignettes on the Mean-Field Analysis of Neural Networks", Workshop on Optimal Transport, Mean-Field Models, and Machine Learning, Institute for Advanced Studies (IAS), Technical University of Munich (TUM), April 2023.
"Fundamental Limits of Two-layer Autoencoders, and Achieving Them with Gradient Methods", Information Theory and Applications (ITA) Workshop, UCSD, San Diego, February 2023.
"Inference in High Dimensions for (Mixed) Generalized Linear Models: the Linear, the Spectral and the Approximate", Information Theory and Data Science Workshop, National University of Singapore (NUS), January 2023.
"Understanding gradient descent for over-parameterized deep neural networks: Insights from mean-field theory and the neural tangent kernel", Colloquium of the Department of Mathematics, Technical University of Munich (TUM), November 2022.
"Inference in High Dimensions for (Mixed) Generalized Linear Models: the Linear, the Spectral and the Approximate", Stochastics and Statistics Seminar, MIT, November 2022.
"Gradient Descent for Deep Neural Networks: New Perspectives from Mean-field and NTK", University of Pennsylvania, November 2022.
"Inference in High Dimensions for Generalized Linear Models: the Linear, the Spectral and the Approximate", Canadian Workshop on Information Theory (CWIT), Ottawa, June 2022.
"Mean-field Analysis of Piecewise Linear Solutions for Wide ReLU Networks", Information Theory and Applications (ITA) Workshop, UCSD, San Diego, May 2022.
"Understanding Gradient Descent for Over-parameterized Deep Neural Networks", EPFL, April 2022.
—, ASU LIONS Seminar Series, Arizona State University, March 2022 (Online).
"Gradient Descent for Deep Neural Networks: New Perspectives from Mean-field and NTK", Math ML seminar MPI MiS + UCLA, MPI Leipzig, March 2022.
"Landscape Connectivity in Deep Neural Networks: Mean-field and Beyond", Loss Landscape of Neural Networks: theoretical insights and practical implications, EPFL, February 2022 (Online).
"Tight Bounds on the Smallest Eigenvalue of the Neural Tangent Kernel for Deep Neural Networks", DeepMind, December 2021 (Online).
"Analysis of a Two-Layer Neural Network via Displacement Convexity", Geometric Methods in Optimization and Sampling, Working Group: Mean Field NN, Simons Institute for the Theory of Computing, Berkeley, October 2021 (Online).
—, Theory of Neural Nets Seminar, EPFL, June 2021 (Online).
"Inference in High Dimensions for Generalized Linear Models: the Linear, the Spectral and the Approximate", ISOR Colloquium, University of Vienna, May 2021 (Online).
"Mode Connectivity and Convergence of Gradient Descent for (Not So) Over-parameterized Deep Neural Networks", International School for Advanced Studies (SISSA), March 2021 (Online).
"Understanding Gradient Descent for Over-parameterized Deep Neural Networks", Mathematics of Data Seminar, Max Planck Institute for Mathematics in the Sciences (MPI MiS), Leipzig, August 2020 (Online).
—, Youth in High Dimensions, International Centre for Theoretical Physics (ICTP), July 2020 (Online).
—, Technical University of Munich (TUM), June 2020 (Online).
"Landscape Connectivity and Dropout Stability of SGD Solutions for Over-parameterized Neural Networks", Information Theory and Applications (ITA) Workshop, UCSD, San Diego, February 2020.
"Analysis of a Two-Layer Neural Network via Displacement Convexity", Foundations of Data Science Reunion, Simons Institute for the Theory of Computing, Berkeley, December 2019.
—, Deep Learning Seminar, University of Vienna, October 2019.