.. Atsushi Shibagaki documentation master file, created by sphinx-quickstart on Sat Jun 11 10:51:18 2016. You can adapt this file completely to your liking, but it should at least contain the root `toctree` directive. .. role:: underline :class: underline .. role:: pubs :class: pubs .. raw:: html
Atsushi Shibagaki (柴垣篤志) ========================================== .. I am currently a master's student (department of scientific and engineering simulation) in `Takeuchi lab `_, `Nagoya Institute of Technology `_, Japan. I found employment. .. raw:: html
shibagaki.a.mllab.nit _at_ gmail.com
GitHub
Preprint ^^^^^^^^^^ .. raw:: html
    Abstract: We study primal-dual type stochastic optimization algorithms with non-uniform sampling. Our main theoretical contribution in this paper is to present a convergence analysis of Stochastic Primal Dual Coordinate (SPDC) Method with arbitrary sampling. Based on this theoretical framework, we propose Optimality Violation-based Sampling SPDC (ovsSPDC), a non-uniform sampling method based on Optimality Violation. We also propose two efficient heuristic variants of ovsSPDC called ovsSDPC+ and ovsSDPC++. Through intensive numerical experiments, we demonstrate that the proposed method and its variants are faster than other state-of-the-art primal-dual type stochastic optimization methods.

Publications ^^^^^^^^^^^^^^ .. :pubs:`Preprint` :pubs:`International Conferences (refereed)` .. raw:: html
    Abstract: We study large-scale machine learning problems in changing environments where a small part of the dataset is modified, and the effect of the data modification must be monitored in order to know how much the modification changes the optimal model. When the entire dataset is large, even if the amount of the data modification is fairly small, the computational cost for re-training the model would be prohibitively large. In this paper, we propose a novel method, called the optimal solution bounding (OSB), for monitoring such a data modification effect on the optimal model by efficiently evaluating (without actually re-training) it. The proposed method provides bounds on the unknown optimal model with the cost proportional only to the size of the data modification.

    Abstract: The problem of learning a sparse model is conceptually interpreted as the process of identifying active features/samples and then optimizing the model over them. Recently introduced safe screening allows us to identify a part of nonactive features/samples. So far, safe screening has been individually studied either for feature screening or for sample screening. In this paper, we introduce a new approach for safely screening features and samples simultaneously by alternatively iterating feature and sample screening steps. A significant advantage of considering them simultaneously rather than individually is that they have a synergy effect in the sense that the results of the previous safe feature screening can be exploited for improving the next safe sample screening performances, and viceversa. We first theoretically investigate the synergy effect, and then illustrate the practical advantage through intensive numerical experiments for problems with large numbers of features and samples.

    Abstract: Careful tuning of a regularization parameter is indispensable in many machine learning tasks because it has a significant impact on generalization performances. Nevertheless, current practice of regularization parameter tuning is more of an art than a science, e.g., it is hard to tell how many grid-points would be needed in cross-validation (CV) for obtaining a solution with sufficiently small CV error. In this paper we propose a novel framework for computing a lower bound of the CV errors as a function of the regularization parameter, which we call regularization path of CV error lower bounds. The proposed framework can be used for providing a theoretical approximation guarantee on a set of solutions in the sense that how far the CV error of the current best solution could be away from best possible CV error in the entire range of the regularization parameters. Our numerical experiments demonstrate that a theoretically guaranteed choice of a regularization parameter in the above sense is possible with reasonable computational costs.

:pubs:`Domestic Conferences (not refereed)` .. raw:: html
変数の保持と削除に関するセーフルールによるスパースモデル最適化問題のサイズ縮小と高速化に関する一考察
烏山昌幸, 柴垣篤志, 竹内一郎
信学技報, vol. 116, no. 500, IBISML 2016-107, pp. 57-62, 東京, 2017年3月.
paper
経験損失最小化問題における高速感度分析に関する一提案
花田博幸, 柴垣篤志, 佐久間淳, 竹内一郎
信学技報, vol. 116, no. 209, IBISML 2016-35, pp. 203-210, 富山, 2016年9月.
paper
スパースモデルのための特徴と標本の同時セーフスクリーニング
柴垣篤志, 烏山昌幸, 畑埜晃平, 竹内一郎
信学技報, vol. 116, no. 121, IBISML 2016-4, pp. 201-208, 沖縄, 2016年7月.
paper
L2正則化凸損失関数最小化問題のための検証誤差近似保証付きモデル選択
柴垣篤志, 鈴木良規, 竹内一郎
信学技報, vol. 114, no. 502, IBISML 2014-96, pp. 79-86, 京都, 2015年3月.
paper
Activities ^^^^^^^^^^^^^ .. rst-class:: actli - **Mar (2017):** Got master's degree form Nagoya Institute of Technology (Department of Scientific and Engineering Simulation) - **Dec (2016):** Poster presentation in `情報系 WINTER FESTA Episode2 `_ @ Tokyo - **Nov (2016):** Poster presentation in `IBIS2016 `_ @ Kyoto Univ. - **Jul (2016):** Oral presentation in `IBISML25 `_ @ Okinawa. - **Jun (2016):** Poster presentation in `ICML 2016 `_ @ New York City - **Dec (2015):** Poster presentation in `NIPS 2015 `_ @ Montreal - **Nov (2015):** Poster presentation in `IBIS2015 `_ @ Tsukuba - **Aug - Sep (2015):** Poster presentation in `Machine Learning Summer School (MLSS) 2015 `_ @ Kyoto Univ. - **Mar (2015):** Got bachelor's degree form Nagoya Institute of Technology (Department of Computer Science) **Award:** `電影会賞 `_ - **Mar (2015):** Oral presentation in `IBISML20 `_ @ Kyoto Univ. Memo ===== .. toctree:: :titlesonly: metriclearning noisylabel ssl representation other proceedings ltr agri