Description
SAGA is a method in the spirit of SAG, SDCA, MISO and SVRG, a set of recently proposed incremental gradient algorithms with fast linear convergence rates. SAGA improves on the theory behind SAG and SVRG, with better theoretical convergence rates, and has support for composite objectives where a proximal operator is used on the regulariser. Unlike SDCA, SAGA supports non-strongly convex problems directly, and is adaptive to any inherent strong convexity of the problem.
Papers Using This Method
Rethinking Verification for LLM Code Generation: From Generation to Testing2025-07-09ATMM-SAGA: Alternating Training for Multi-Module with Score-Aware Gated Attention SASV system2025-05-23A digital perspective on the role of a stemma in material-philological transmission studies2025-05-11SAGA: A Security Architecture for Governing AI Agentic Systems2025-04-27SAGA: Semantic-Aware Gray color Augmentation for Visible-to-Thermal Domain Adaptation across Multi-View Drone and Ground-Based Vision Systems2025-04-22Variance-Reduced Fast Operator Splitting Methods for Stochastic Generalized Equations2025-04-17SAGA: Surface-Aligned Gaussian Avatar2024-12-01On the SAGA algorithm with decreasing step2024-10-02SAGA: Synthesis Augmentation with Genetic Algorithms for In-Memory Sequence Optimization2024-06-14Variance-Reduced Fast Krasnoselkii-Mann Methods for Finite-Sum Root-Finding Problems2024-06-04SPABA: A Single-Loop and Probabilistic Stochastic Bilevel Algorithm Achieving Optimal Sample Complexity2024-05-29Segment Any 3D Gaussians2023-12-01PROMISE: Preconditioned Stochastic Optimization Methods by Incorporating Scalable Curvature Estimates2023-09-05Variance reduction techniques for stochastic proximal point algorithms2023-08-18Growing and Serving Large Open-domain Knowledge Graphs2023-05-16Attacking the Spike: On the Transferability and Security of Spiking Neural Networks to Adversarial Examples2022-09-07Tackling Data Heterogeneity: A New Unified Framework for Decentralized SGD with Sample-induced Topology2022-07-08An Adaptive Incremental Gradient Method With Support for Non-Euclidean Norms2022-04-28Saga: A Platform for Continuous Construction and Serving of Knowledge At Scale2022-04-15A framework for bilevel optimization that enables stochastic and global variance reduction algorithms2022-01-31