BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CREST - ECPv5.1.3//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:CREST
X-ORIGINAL-URL:https://crest.science
X-WR-CALDESC:Events for CREST
BEGIN:VTIMEZONE
TZID:Europe/Helsinki
BEGIN:DAYLIGHT
TZOFFSETFROM:+0200
TZOFFSETTO:+0300
TZNAME:EEST
DTSTART:20260329T010000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0300
TZOFFSETTO:+0200
TZNAME:EET
DTSTART:20261025T010000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Europe/Helsinki:20260309T133000
DTEND;TZID=Europe/Helsinki:20260316T173000
DTSTAMP:20260709T203928
CREATED:20260212T083208Z
LAST-MODIFIED:20260212T083217Z
UID:18805-1773063000-1773682200@crest.science
SUMMARY:Xunyu Zhou (Columbia University) - Introduction to Continuous-Time Reinforcement Learning
DESCRIPTION:INTRODUCTION TO CONTINUOUS-TIME REINFORCEMENT LEARNING\nXUNYU ZHOU\nColumbia University\nIndustrial Engineering & Operations Research Department \nSchedule: \nMarch 9\, 2026 | 1:30PM – 5:30PM – Room 2006\nMarch 12\, 2026 | 1:30PM – 5:30PM – Room 2006\nMarch 16\, 2026 | 1:30PM – 5:30PM – Room 2006 \nContent:\nThis crash course covers fundamental theory and algorithms for reinforcement learning with continuous-time controlled diffusion processes\, which have been developed in the last five years. It includes the following topics.\n1. Exploration vs exploitation: relaxed control\, entropy regularization\, exploratory HJB equation and Gibbs measure.\n2. Gaussian exploration under linear-quadratic control: optimality of Gaussian exploration and cost of exploration.\n3. Temperature control of Langevin diffusions: simulated annealing for nonconvex optimization and optimal temperature control.\n4. Policy evaluation: martingale characterization\, martingale loss function and martingale orthogonality conditions.\n5. Policy gradient: policy gradient via policy evaluation\, and temporal difference learning. q-learning theory: generalized Hamiltonian and policy improvement\, Q-function and q-function\, martingale characterization of q-function. \nEvaluation: \nTake home project. \n
URL:https://crest.science/event/xunyu-zhou-columbia-university-introduction-to-continuous-time-reinforcement-learning/
CATEGORIES:Doctoral Courses,Finance
END:VEVENT
BEGIN:VEVENT
DTSTART;VALUE=DATE:20260316
DTEND;VALUE=DATE:20260331
DTSTAMP:20260709T203928
CREATED:20251210T123817Z
LAST-MODIFIED:20260324T072239Z
UID:18644-1773619200-1774915199@crest.science
SUMMARY:Taiji Suzuki (Univ. Tokyo) - Feature learning perspective of deep foundation models: statistical and optimization theories
DESCRIPTION:FEATURE LEARNING PERSPECTIVE OF DEEP FOUNDATION MODELS: STATISTICAL AND OPTIMIZATION THEORIES\nTAIJI SUZUKI\nUniversity of Tokyo\nDepartment of Mathematical Informatics \nSchedule: \nMarch 16\, 2026 | 2PM – 5PM – Room 2005\nMarch 19\, 2026 | 2PM – 5PM – Room 2005\nMarch 26\, 2026 | 2PM – 5PM – Room 2005\nMarch 30\, 2026 | 2PM – 5PM – Room 2005 \nContent:\nThe main focus of this lecture will be on the feature learning aspects of deep foundation models\, especially about the benefit of feature learning to achieve better predictive accuracy and more efficient optimization. As the deep foundation models are developed following the scaling law\, a theoretical understanding of the learning principles behind practice is getting more important. For superior generalization of deep models\, it is essentially important to acquire compressed representations avoiding mere memorization\, making representation/feature learning fundamental. It has been theoretically shown that deep learning gains various advantages in generalization via its feature learning ability which naturally arises from its deep structure. It will be discussed how the feature learning ability affects the rate of convergence by comparing it with the sub-optimal rate of non-feature learning methods. This issue will be discussed not only from statistical  perspective but also from optimization perspective. Interesting examples include estimating a Gaussian single index model in which the computational complexity can be characterized by quantities socalled information exponent and generative exponent. If time permits\, optimization guarantees by mean field Langevin dynamics and its statistical property will also be discussed. Furthermore\, feature learning is significant not only during pre-training but also during test-time inference. This will be demonstrated concisely using in-context learning as an example. It will be discussed how the test time feature learning as well as pre-training feature learning affects the performance of test time inference. In summary\, the following topics will be covered in the lecture (but some of them could be omitted depending on time constraint):\n• Nonparametric function estimation by deep learning on high dimension data and its minimax optimality.\n• Stochastic gradient descent for neural network training; Gaussian single index model\, k-parity problem\, information exponent\, CSQ/SQ lower bound.\n• Mean field Langevin dynamics and its statistical property.\n• Test time inference and test time scaling: Transformer\, in-context learning\, chain-ofthought. \nEvaluation: \nTake home project. \nReferences:\nE. Giné and R. Nickl. Mathematical foundations of infinite-dimensional statistical models. Cambridge Series in Statistical and Probabilistic Mathematics. Cambridge University Press\, 2015.\nS. Hayakawa and T. Suzuki. On the minimax optimality and superiority of deep neural network learning over sparse parameter spaces. Neural Networks\, 123:343–361\, 2020.\nJ. Kim\, T. Nakamaki\, and T. Suzuki. Transformers are minimax optimal nonparametric incontext learners. In Advances in Neural Information Processing Systems\, volume 37\, pages 106667–106713\, 2024.\nJ. D. Lee\, K. Oko\, T. Suzuki\, and D. Wu. Neural network learns low-dimensional polynomials with SGD near the information-theoretic limit. In Advances in Neural Information Processing Systems\, volume 37\, pages 58716–58756\, 2024.\nN. Nishikawa\, Y. Song\, K. Oko\, D. Wu\, and T. Suzuki. Nonlinear transformers can perform inference-time feature learning. In Proceedings of the 42nd International Conference on Machine Learning\, volume 267 of Proceedings of Machine Learning Research\, pages 46554– 46585. PMLR\, 2025.\nA. Nitanda\, D. Wu\, and T. Suzuki. Convex analysis of the mean field Langevin dynamics. In Proceedings of The 25th International Conference on Artificial Intelligence and Statistics\, volume 151 of Proceedings of Machine Learning Research\, pages 9741–9757. PMLR\, 28–30 Mar 2022.\nNitanda\, A. Lee\, D. T. X. Kai\, M. Sakaguchi\, and T. Suzuki. Propagation of chaos for mean-field Langevin dynamics and its application to model ensemble. In Forty-second International Conference on Machine Learning\, 2025.\nK. Oko\, Y. Song\, T. Suzuki\, and D. Wu. Pretrained transformer efficiently learns lowdimensional target functions in-context. In Advances in Neural Information Processing Systems\, volume 37\, pages 77316–77365\, 2024.\nT. Suzuki. Adaptivity of deep ReLU network for learning in Besov and mixed smooth Besov spaces: optimal rate and curse of dimensionality. In International Conference on Learning Representations\, 2019.\nT. Suzuki and A. Nitanda. Deep learning is adaptive to intrinsic dimensionality of model smoothness in anisotropic Besov space. In Advances in Neural Information Processing Systems\, volume 34\, pages 3609–3621\, 2021. A. B. Tsybakov. \n  \n
URL:https://crest.science/event/taiji-suzuki-univ-tokyo-feature-learning-perspective-of-deep-foundation-models-statistical-and-optimization-theories/
CATEGORIES:Doctoral Courses,Statistics
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Helsinki:20260316T121500
DTEND;TZID=Europe/Helsinki:20260316T133000
DTSTAMP:20260709T203928
CREATED:20260302T133156Z
LAST-MODIFIED:20260305T104626Z
UID:18847-1773663300-1773667800@crest.science
SUMMARY:Domenico GIANNONE (Johns Hopkins University) "Bayesian Inference in IV Regression"
DESCRIPTION:[vc_row][vc_column][vc_column_text]Macro seminar\nTime : 12h15 – 13h30 \nDate : 16th  March 2026 \nSalle 3001 \nDomenico GIANNONE (Johns Hopkins University) “Bayesian Inference in IV Regression” \nAbstract: It is well known that standard frequentist inference breaks down in IV regressions with weak instruments. Bayesian inference with diffuse priors suffers from the same problem. We show that the issue arises because flat priors on the first-stage coefficients overstate instrument strength. In contrast\, inference improves drastically when an uninformative prior is specified directly on the concentration parameter—the key nuisance parameter capturing instrument relevance. The  resulting Bayesian credible intervals are asymptotically equivalent to the frequentist confidence intervals based on conditioning approaches\, and remain robust to weak instruments.\n \nJoint work : Michele Lenza and Giorgio Primiceri \nOrganizer :  Alessandro RIBONI \n
URL:https://crest.science/event/domenico-giannone-johns-hopkins-university-bayesian-inference-in-iv-regression/
CATEGORIES:Macroeconomics,Seminars
ATTACH;FMTTYPE=:
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Europe/Helsinki:20260316T140000
DTEND;TZID=Europe/Helsinki:20260316T153000
DTSTAMP:20260709T203928
CREATED:20260223T091638Z
LAST-MODIFIED:20260310T121344Z
UID:18818-1773669600-1773675000@crest.science
SUMMARY:Gabor LUGOSI (Universitat Pompeu Fabra) - Structure learning and property testing in graphical models
DESCRIPTION:Statistical Seminar: Every Monday at 2:00 pm.\nTime: 2:00 pm – 3:00 pm\nDate: 16th March\nPlace: 3001 \n  \nGabor LUGOSI (Universitat Pompeu Fabra) – Structure learning and property testing in graphical models \n  \n Abstract:  \nThe dependence structure of high-dimensional distributions is often modeled by graphical models. The problem of learning the graph underlying such distributions has received a lot of attention in statistics and machine learning. In problems of very high dimension\, it is often too costly even to store the sample covariance matrix. We propose a model in which one can query single entries of the covariance matrix. We construct efficient algorithms for structure recovery in Gaussian graphical models with query complexity that is quasi-linear in the dimension. We present algorithms that work for trees and\, more generally\, for graphs of small treewidth. We also discuss hypothesis testing of properties of the underlying graph.\nThe talk is based on joint work with Sofiya Burova\, Francisco Calvillo\, Luc Devroye\, Jakub Truszkowski\, Vasiliki Velona\, and Piotr Zwiernik. \n  \nOrganizers: \nAnna KORBA (CREST)\, Vincent DIVOL (CREST) \, Jaouad MOURTADA (CREST) \nSponsors:\nCREST-CMAP \n
URL:https://crest.science/event/gabor-lugosi-universitat-pompeu-fabra-tba/
CATEGORIES:Seminars,Statistics
ATTACH;FMTTYPE=:
END:VEVENT
END:VCALENDAR