1 L 1 Bregman Iterative Shrinkage Threshold Algorithm
実装例1: L 1正則化問題 [部分問題] Bregman距離に を用いると Iterative Shrinkage Threshold Algorithm (ISTA)
高速化手法1 [Nesterov, 1983], [Beck, Teboulle, 2009] を更新 ISTA 高速化 Fast ISTA (FISTA) Beck and Teboulle, A fast iterative shrinkage-thresholding algorithm for linear inverse problems SIAM Journal on Imaging Sciences, 2009.
高速化手法2 [Nesterov, 1988], [Auslender, Teboulle, 2006] を更新
高速化手法3 [Nesterov 2005], [Tseng, 2008] を更新
アルゴリズムの関係 近接勾配法 ISTA Pが標示関数 近接点法 射影勾配法 Mirror Descent 法 EG法 高速化法 FISTA 高速化EG法 X. Zhang, A. Saha, S. V. N. Vishwanathan, “Accelerated Training of Max-Margin Markov Networks with Kernels”, Algorithmic Learning Theory (ALT), 2011
(最適化の分野の) サーベイ文献+α [近接勾配法] Paul Tseng, Approximation accuracy, gradient methods, and error bound for structured convex optimization, Mathematical Programming, Ser, B, 125, pp. 263 -296, 2010. [エラーバウンド性] Jong-Shi Pang, Error bounds in mathematical programming, Mathematical Programming , 79, pp. 299 -332, 1997. [最新論文] http: //www. optimization-online. org/
- Slides: 36