In this paper, we consider approximate Frank-Wolfe (FW) algorithms to solve convex optimization problems over graph-structured support sets where the linear minimization oracle (LMO) cannot be efficiently obtained in general. We first demonstrate that two popular approximation assumptions (additive and multiplicative gap errors) are not applicable in that no cheap gap-approximate LMO oracle exists. Thus, approximate dual maximization oracles (DMO) are proposed, which approximate the inner product rather than the gap. We prove that the standard FW method using a $\delta$-approximate DMO converges as $\mathcal{O}((1-\delta) \sqrt{s}/\delta)$ in the worst case, and as $\mathcal{O}(L/(\delta^2 t))$ over a $\delta$-relaxation of the constraint set. Furthermore, when the solution is on the boundary, a variant of FW converges as $\mathcal{O}(1/t^2)$ under the quadratic growth assumption. Our empirical results suggest that even these improved bounds are pessimistic, showing fast convergence in recovering real-world images with graph-structured sparsity.
翻译:在本文中,我们考虑大约Frank-Wolfe (FW) 算法,以解决在平面结构支持组群中无法有效获得线性最小化或电极(LMO)的混凝土优化问题。我们首先表明,在最坏的情况下,两种流行的近似假设(额外和倍增差差差差)并不适用,因为不存在任何廉价的差幅-接近LMO或电极差差幅。因此,提出了接近内部产品而不是差距的近似双倍最大化或电法(DMO)。此外,在边界上找到解决办法时,使用美元/delta$-ap 近似DMO的标准FW 方法将美元(1- delta)\ mathcal{O} ((1-\delta)\ sqrt{s}/\delta) 美元汇合为美元,在最坏的情况下,以美元/(delta2) 美元/telta2 t(DMO) 和美元以上限制值调和美元。此外,在边界上找到一个FW的变式方法的折合为$\ mathcaladal{O}O} (1/2) 在平面图中显示,我们的正平面图状图中,我们正在呈现着快速增长。