Value-Function-based Sequential Minimization for Bi-level Optimization
This work addresses a foundational problem in machine learning for researchers and practitioners dealing with bi-level optimization, offering a more general and efficient solution, though it is incremental in improving existing methods.
The paper tackles the limitations of gradient-based bi-level optimization methods, which rely on restrictive assumptions and are computationally inefficient for high-dimensional tasks, by introducing BVFSM, a new algorithm that avoids costly gradient and Hessian inverse calculations and extends to challenging scenarios like pessimistic BLO, achieving asymptotic convergence without convexity assumptions and demonstrating superiority in experiments.
Gradient-based Bi-Level Optimization (BLO) methods have been widely applied to handle modern learning tasks. However, most existing strategies are theoretically designed based on restrictive assumptions (e.g., convexity of the lower-level sub-problem), and computationally not applicable for high-dimensional tasks. Moreover, there are almost no gradient-based methods able to solve BLO in those challenging scenarios, such as BLO with functional constraints and pessimistic BLO. In this work, by reformulating BLO into approximated single-level problems, we provide a new algorithm, named Bi-level Value-Function-based Sequential Minimization (BVFSM), to address the above issues. Specifically, BVFSM constructs a series of value-function-based approximations, and thus avoids repeated calculations of recurrent gradient and Hessian inverse required by existing approaches, time-consuming especially for high-dimensional tasks. We also extend BVFSM to address BLO with additional functional constraints. More importantly, BVFSM can be used for the challenging pessimistic BLO, which has never been properly solved before. In theory, we prove the asymptotic convergence of BVFSM on these types of BLO, in which the restrictive lower-level convexity assumption is discarded. To our best knowledge, this is the first gradient-based algorithm that can solve different kinds of BLO (e.g., optimistic, pessimistic, and with constraints) with solid convergence guarantees. Extensive experiments verify the theoretical investigations and demonstrate our superiority on various real-world applications.