More Like this

1. On The Complexity of First-Order Methods in Stochastic Bilevel Optimization

   Kwon, Jeongyeol; Kwon, Dohyun; Lyu, Hanbaek (July 2024, Proceedings of Machine Learning Research)

   We consider the problem of finding stationary points in Bilevel optimization when the lower-level problem is unconstrained and strongly convex. The problem has been extensively studied in recent years; the main technical challenge is to keep track of lower-level solutions $y^*(x)$ in response to the changes in the upper-level variables $$x$$. Subsequently, all existing approaches tie their analyses to a genie algorithm that knows lower-level solutions and, therefore, need not query any points far from them. We consider a dual question to such approaches: suppose we have an oracle, which we call $y^*$-aware, that returns an $$O(\epsilon)$$-estimate of the lower-level solution, in addition to first-order gradient estimators {\it locally unbiased} within the $$\Theta(\epsilon)$$-ball around $y^*(x)$. We study the complexity of finding stationary points with such an $y^*$-aware oracle: we propose a simple first-order method that converges to an $$\epsilon$$ stationary point using $$O(\epsilon^{-6}), O(\epsilon^{-4})$$ access to first-order $y^*$-aware oracles. Our upper bounds also apply to standard unbiased first-order oracles, improving the best-known complexity of first-order methods by $$O(\epsilon)$$ with minimal assumptions. We then provide the matching $$\Omega(\epsilon^{-6})$$, $$\Omega(\epsilon^{-4})$$ lower bounds without and with an additional smoothness assumption on $y^*$-aware oracles, respectively. Our results imply that any approach that simulates an algorithm with an $y^*$-aware oracle must suffer the same lower bounds. 

   more » « less
2. A Near-Optimal Algorithm for Stochastic Bilevel Optimization viaDouble-Momentum

   Prashant Khanduri, Siliang Zeng (October 2021, Advances in neural information processing systems)

   This work proposes a new algorithm – the Single-timescale Double-momentum Stochastic Approximation (SUSTAIN) –for tackling stochastic unconstrained bilevel optimization problems. We focus on bilevel problems where the lower level subproblem is strongly-convex and the upper level objective function is smooth. Unlike prior works which rely on two-timescale or double loop techniques, we design a stochastic momentum-assisted gradient estimator for both the upper and lower level updates. The latter allows us to control the error in the stochastic gradient updates due to inaccurate solution to both subproblems. If the upper objective function is smooth but possibly non-convex, we show that SUSTAIN requires $${O}(\epsilon^{-3/2})$$ iterations (each using $O(1)$ samples) to find an $$\epsilon$$-stationary solution. The $$\epsilon$$-stationary solution is defined as the point whose squared norm of the gradient of the outer function is less than or equal to $$\epsilon$$. The total number of stochastic gradient samples required for the upper and lower level objective functions match the best-known complexity for single-level stochastic gradient algorithms. We also analyze the case when the upper level objective function is strongly-convex. 

   more » « less
3. On The Complexity of First-Order Methods in Stochastic Bilevel Optimization

   Kwon, Jeongyeol; Kwon, Dohyun; Lyu, Hanbaek (July 2024, Proceedings of Machine Learning Research)

   We consider the problem of finding stationary points in Bilevel optimization when the lower-level problem is unconstrained and strongly convex. The problem has been extensively studied in recent years; the main technical challenge is to keep track of lower-level solutions $y^*(x)$ in response to the changes in the upper-level variables $$x$$. Subsequently, all existing approaches tie their analyses to a genie algorithm that knows lower-level solutions and, therefore, need not query any points far from them. We consider a dual question to such approaches: suppose we have an oracle, which we call $y^*$-aware, that returns an $$O(\epsilon)$$-estimate of the lower-level solution, in addition to first-order gradient estimators locally unbiased within the $$\Theta(\epsilon)$$-ball around $y^*(x)$. We study the complexity of finding stationary points with such an $y^*$-aware oracle: we propose a simple first-order method that converges to an $$\epsilon$$ stationary point using $$O(\epsilon^{-6}), O(\epsilon^{-4})$$ access to first-order $y^*$-aware oracles. Our upper bounds also apply to standard unbiased first-order oracles, improving the best-known complexity of first-order methods by $$O(\epsilon)$$ with minimal assumptions. We then provide the matching $$\Omega(\epsilon^{-6})$$, $$\Omega(\epsilon^{-4})$$ lower bounds without and with an additional smoothness assumption, respectively. Our results imply that any approach that simulates an algorithm with an $y^*$-aware oracle must suffer the same lower bounds. 

   more » « less
4. Achieving O(\epsilon^{-1.5}) Complexity in Hessian/Jacobian-free Stochastic Bilevel Optimization

   Yang, Y; Xiao, P; Ji, K (February 2024, Advances in Neural Information Processing Systems)

   In this paper, we revisit the bilevel optimization problem, in which the upper-level objective function is generally nonconvex and the lower-level objective function is strongly convex. Although this type of problem has been studied extensively, it still remains an open question how to achieve an $$\mathcal{O}(\epsilon^{-1.5})$$ sample complexity in Hessian/Jacobian-free stochastic bilevel optimization without any second-order derivative computation. To fill this gap, we propose a novel Hessian/Jacobian-free bilevel optimizer named FdeHBO, which features a simple fully single-loop structure, a projection-aided finite-difference Hessian/Jacobian-vector approximation, and momentum-based updates. Theoretically, we show that FdeHBO requires $$\mathcal{O}(\epsilon^{-1.5})$$ iterations (each using $$\mathcal{O}(1)$$ samples and only first-order gradient information) to find an $$\epsilon$$-accurate stationary point. As far as we know, this is the first Hessian/Jacobian-free method with an $$\mathcal{O}(\epsilon^{-1.5})$$ sample complexity for nonconvex-strongly-convex stochastic bilevel optimization. 

   more » « less
5. A nonconvex formulation for low rank subspace clustering: algorithms and convergence analysis

   Jiang, Hao; Robinson, Daniel P.; Vidal, Rene; You, Chong (April 2018, Computational optimization and applications)

   We consider the problem of subspace clustering with data that is potentially corrupted by both dense noise and sparse gross errors. In particular, we study a recently proposed low rank subspace clustering approach based on a nonconvex modeling formulation. This formulation includes a nonconvex spectral function in the objective function that makes the optimization task challenging, e.g., it is unknown whether the alternating direction method of multipliers (ADMM) framework proposed to solve the nonconvex model formulation is provably convergent. In this paper, we establish that the spectral function is differentiable and give a formula for computing the derivative. Moreover, we show that the derivative of the spectral function is Lipschitz continuous and provide an explicit value for the Lipschitz constant. These facts are then used to provide a lower bound for how the penalty parameter in the ADMM method should be chosen. As long as the penalty parameter is chosen according to this bound, we show that the ADMM algorithm computes iterates that have a limit point satisfying first-order optimality conditions. We also present a second strategy for solving the nonconvex problem that is based on proximal gradient calculations. The convergence and performance of the algorithms is verified through experiments on real data from face and digit clustering and motion segmentation. 

   more » « less