Conditional Gradient Methods From Core Principles to AI Applications

Conditional Gradient Methods: From Core Principles to AI Applications offers a definitive and modern treatment of one of the most elegant and versatile algorithmic families in optimization: the Frank–Wolfe method and its many variants. Originally proposed in the 1950s, these projection-free techniques have seen a powerful resurgence, now playing a central role in machine learning, signal processing, and large-scale data science. This comprehensive monograph unites deep theoretical insights with practical considerations, guiding readers through the foundations of constrained optimization and into cutting-edge territory, including stochastic, online, and distributed settings. With a clear narrative, rigorous proofs, and illuminating illustrations, the book demystifies adaptive variants, away-steps, and the nuances of dealing with structured convex sets. Implementations of most of the algorithms in the book are available on a supplementary website or in the FrankWolfe.jl Julia package.

Gábor Braun is a member of the Zuse Institute Berlin.

Alejandro Carderera is a senior applied researcher at GitHub, working in Copilot's Applied Science and Models team, focusing on code completions and code review.

Cyrille W. Combettes is a quantitative researcher at Capital Fund Management in Paris.

Hamed Hassani is an associate professor in the Department of Electrical and Systems Engineering, the Department of Computer and Information Systems, and the Department of Statistics and Data Science at the University of Pennsylvania.

Aryan Mokhtari is an assistant professor in the Department of Electrical and Computer Engineering at the University of Texas at Austin, where he holds the Fellow of Texas Instruments/Kilby.