Abstract Gradient Training: A Unified Certification Framework for Data Poisoning, Unlearning [...] @FieldsInstitute
Abstract Gradient Training: A Unified Certification Framework for Data Poisoning, Unlearning [...]  @FieldsInstitute
Uploaded October 2026 | Updated November 2026, 1 day ago
Full Title: Abstract Gradient Training: A Unified Certification Framework for Data Poisoning, Unlearning, and Differential Privacy

Speaker: Philip Sosnin, Imperial College London
Date: October 9, 2026
Abstract: fields.utoronto.ca/talks/Abstract-Gradient-Training-Unified-Certification-Framework-Data-Poisoning-Unlearning-and

2026-2027 Mathematical AI Seminar: fields.utoronto.ca/activities/26-27/mathai

Centre for Mathematical AI: fields.utoronto.ca/centres/centre-mathematical-ai
Abstract Gradient Training: A Unified Certification Framework for Data Poisoning, Unlearning [...]Optimal transport between laws of random measures: total convexity and the Monge problemOn the q-integrability of the p-Wasserstein barycentersMust you see everything? On impartiality and the power of partial information.Lecture 08 | Mathematical Foundations of AIQuantum Optimal Transport Barycenters: Existence, Duality, and Gaussian RigidityDoing some math with AIIsoenergetic degeneracy generically creates meandering invariant toriQuantum Resource Estimation in PennyLaneAlchemy as a quantum control problem  Theory and ExperimentsShareholder Engagement in an ESG-CAPM with Incomplete Markets: Much ado about nothing?Lecture 03 | Kernel Adaptation: Feature Learning Beyond the Gaussian Process
Fields Institute |

Abstract Gradient Training: A Unified Certification Framework for Data Poisoning, Unlearning [...]

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER