Litcius/Paper detail

Dynamical mean-field theory for stochastic gradient descent in Gaussian mixture classification

Francesca Mignacco, Florent Krząkała, Pierfrancesco Urbani, Lenka Zdeborová

2021HAL (Le Centre pour la Communication Scientifique Directe)41 citationsDOI

Abstract

International audience

Topics & Concepts

Stochastic gradient descentLimit (mathematics)GaussianGradient descentMathematicsGeneralizationApplied mathematicsGaussian processStochastic processStatistical physicsBalanced flowComputer scienceField (mathematics)Mathematical optimizationArtificial neural networkAlgorithmMathematical analysisArtificial intelligencePhysicsStatisticsPure mathematicsQuantum mechanicsStochastic Gradient Optimization TechniquesNeural Networks and ApplicationsMarkov Chains and Monte Carlo Methods
Dynamical mean-field theory for stochastic gradient descent in Gaussian mixture classification | Litcius