Committee |
Date Time |
Place |
Paper Title / Authors |
Abstract |
Paper # |

IBISML |
2022-03-08 10:25 |
Online |
Online |
Robust computation of optimal transport by β-potential regularization Shintaro Nakamura (Univ. Tokyo), Han Bao (Univ.Tokyo/RIKEN), Masashi Sugiyama (RIKEN/Univ. Tokyo) IBISML2021-31 |
Optimal transport (OT) has become a widely used tool to measure the discrepancy between probability distributions
in th... [more] |
IBISML2021-31 pp.8-14 |

IBISML |
2022-03-09 13:30 |
Online |
Online |
Is the Performance of My Deep Network Too Good to Be True?
-- A Direct Approach to Estimating the Bayes Error in Binary Classification -- Takashi Ishida (UTokyo), Ikko Yamane (Université Paris Dauphine-PSL/RIKEN), Nontawat Charoenphakdee (UTokyo), Gang Niu (RIKEN), Masashi Sugiyama (RIKEN/UTokyo) IBISML2021-44 |
[more] |
IBISML2021-44 pp.38-45 |

IBISML |
2021-03-02 10:00 |
Online |
Online |
Learning from Noisy Complementary Labels with Robust Loss Functions Hiroki Ishiguro (UTokyo), Takashi Ishida (UTokyo/RIKEN), Masashi Sugiyama (RIKEN/UTokyo) IBISML2020-34 |
It has been demonstrated that large-scale labeled datasets facilitate the success of machine learning. However, collecti... [more] |
IBISML2020-34 pp.1-8 |

IBISML |
2020-03-11 10:45 |
Kyoto |
Kyoto University (Cancelled but technical report was issued) |
Calibrated Surrogate Maximization of Linear-Fractional Utility in Binary Classification Han Bao (Univ. of Tokyo/RIKEN), Masashi Sugiyama (RIKEN/Univ. of Tokyo) IBISML2019-43 |
Complex classification performance metrics such as the F-measure and Jaccard index are often used to handle class imbala... [more] |
IBISML2019-43 pp.71-78 |

IBISML |
2017-11-09 13:00 |
Tokyo |
Univ. of Tokyo |
Semi-Supervised AUC Optimization based on Positive-Unlabeled Learning Tomoya Sakai, Gang Niu (UTokyo/RIKEN), Masashi Sugiyama (RIKEN/UTokyo) IBISML2017-40 |
Maximizing the area under the receiver operating characteristic curve (AUC) is a standard approach to imbalanced classif... [more] |
IBISML2017-40 pp.39-46 |

IBISML |
2017-11-10 13:00 |
Tokyo |
Univ. of Tokyo |
[Poster Presentation]
Binary Classification from Positive-Confidence Data Takashi Ishida (SMAM/UTokyo/RIKEN), Gang Niu (UTokyo/RIKEN), Masashi Sugiyama (RIKEN/UTokyo) IBISML2017-62 |
Reducing labeling costs in supervised learning is a critical issue in many practical machine learning applications. In ... [more] |
IBISML2017-62 pp.207-214 |

IBISML |
2017-11-10 13:00 |
Tokyo |
Univ. of Tokyo |
[Poster Presentation]
Multi-Task Learning with Positive and Unlabeled Data and Its Application to Mental State Prediction Hirotaka Kaji, Hayato Yamaguchi (Toyota Motor), Masashi Sugiyama (RIKEN/UTokyo) IBISML2017-66 |
In real-world machine learning applications, we are often faced with a situation where only a small number of training s... [more] |
IBISML2017-66 pp.235-242 |

IBISML |
2017-11-10 13:00 |
Tokyo |
Univ. of Tokyo |
Hierarchical Reinforcement Learning Based on Return-Weighted Density Estimation Takayuki Osa (UTokyo/RIKEN), Masashi Sugiyama (RIKEN/UTokyo) IBISML2017-67 |
We propose a hierarchical reinforcement learning (HRL) methods for learning the optimal policy from a multi-modal reward... [more] |
IBISML2017-67 pp.243-249 |

IBISML |
2017-11-10 13:00 |
Tokyo |
Univ. of Tokyo |
Good Arm Identification via Bandit Feedback Hideaki Kano, Junya Honda (UTokyo/RIKEN), Kentaro Sakamaki (UTokyo), Kentaro Matsuura (Johnson & Johnson), Atsuyoshi Nakamura (HU), Masashi Sugiyama (RIKEN/UTokyo) IBISML2017-82 |
In this paper, we consider and discuss a new stochastic multi-armed bandit problem called {em good arm identification} (... [more] |
IBISML2017-82 pp.339-346 |

NC, IPSJ-BIO, IBISML, IPSJ-MPS [detail] |
2017-06-24 10:20 |
Okinawa |
Okinawa Institute of Science and Technology |
Risk Minimization Framework for Multiple Instance Learning from Positive and Unlabeled Bags Han Bao (Univ. of Tokyo), Tomoya Sakai, Issei Sato (Univ. of Tokyo/RIKEN), Masashi Sugiyama (RIKEN/Univ. of Tokyo) IBISML2017-3 |
Multiple instance learning (MIL) is a variation of traditional supervised learning problems where data (referred to as b... [more] |
IBISML2017-3 pp.55-62 |

NC, IPSJ-BIO, IBISML, IPSJ-MPS [detail] |
2017-06-24 10:45 |
Okinawa |
Okinawa Institute of Science and Technology |
Positive-Unlabeled Learning with Non-Negative Risk Estimator Ryuichi Kiryo (Univ. of Tokyo/RIKEN), Gang Niu (Univ. of Tokyo), Masashi Sugiyama (RIKEN/Univ. of Tokyo) IBISML2017-4 |
From only emph{positive}~(P) and emph{unlabeled}~(U) data, a binary classifier can be trained with PU learning, in which... [more] |
IBISML2017-4 pp.63-70 |

NC, IPSJ-BIO, IBISML, IPSJ-MPS [detail] |
2017-06-24 11:10 |
Okinawa |
Okinawa Institute of Science and Technology |
Learning from Complementary Labels Takashi Ishida (SMAM/Univ. of Tokyo), Gang Niu (Univ. of Tokyo), Masashi Sugiyama (RIKEN/Univ. of Tokyo) IBISML2017-5 |
[more] |
IBISML2017-5 pp.71-78 |

NC, IPSJ-BIO, IBISML, IPSJ-MPS [detail] |
2017-06-25 09:30 |
Okinawa |
Okinawa Institute of Science and Technology |
Expectation Propagation for t-Exponential Family Futoshi Futami, Issei Sato (Univ. of Tokyo/RIKEN), Masashi Sugiyama (RIKEN/Univ. of Tokyo) IBISML2017-6 |
Exponential family distributions are highly useful in machine learning since their calculation can be performed efficien... [more] |
IBISML2017-6 pp.179-184 |

NC, IPSJ-BIO, IBISML, IPSJ-MPS [detail] |
2017-06-25 09:55 |
Okinawa |
Okinawa Institute of Science and Technology |
Stochastic Divergence Minimization for Biterm Topic Model Zhenghang Cui (Univ. of Tokyo), Issei Sato (Univ. of Tokyo/RIKEN), Masashi Sugiyama (RIKEN/Univ. of Tokyo) IBISML2017-7 |
Inferring latent topics of collected short texts is useful for understanding its hidden structure and predicting new con... [more] |
IBISML2017-7 pp.185-192 |

IBISML |
2016-11-16 15:00 |
Kyoto |
Kyoto Univ. |
Structural Change Detection in Lithography Systems Yosuke Otsubo (NIKON), Masashi Sugiyama (RIKEN/UTokyo) IBISML2016-45 |
A lithography system, which consists of various mechanical units such as high-precision optical systems, prints circuit ... [more] |
IBISML2016-45 pp.1-8 |

IBISML |
2016-11-16 15:00 |
Kyoto |
Kyoto Univ. |
Robust supervised learning under uncertainty in dataset shift Weihua Hu, Issei Sato (UTokyo), Masashi Sugiyama (RIKEN/UTokyo) IBISML2016-50 |
When machine learning is deployed in the real world, its performance can be significantly undermined because test data m... [more] |
IBISML2016-50 pp.37-44 |

IBISML |
2016-11-16 15:00 |
Kyoto |
Kyoto Univ. |
Policy Search with High-dimensional Context Variables Voot Tangkaratt (The Univ. of Tokyo), Herke van Hoof, Simone Parisi, Gerhard Neumann, Jan Peters (Technical Univ. of Darmstadt), Masashi Sugiyama (The Univ. of Tokyo) IBISML2016-63 |
[more] |
IBISML2016-63 pp.123-130 |

IBISML |
2016-11-17 14:00 |
Kyoto |
Kyoto Univ. |
Semi-Supervised Classification Based on Classification from Positive and Unlabeled Data Tomoya Sakai, Marthinus Christoffel du Plessis, Gang Niu (UTokyo), Masashi Sugiyama (RIKEN/UTokyo) IBISML2016-80 |
Most of the semi-supervised learning methods developed so far use unlabeled data for regularization purposes under parti... [more] |
IBISML2016-80 pp.243-250 |

IBISML |
2016-03-17 13:50 |
Tokyo |
Institute of Statistical Mathematics |
Least-Squares Log-Density Gradient Clustering for Riemannian Manifolds Mina Ashizawa (UTokyo), Hiroaki Sasaki (NAIST), Tomoya Sakai, Masashi Sugiyama (UTokyo) IBISML2015-96 |
[more] |
IBISML2015-96 pp.17-24 |

IBISML |
2015-11-26 15:00 |
Ibaraki |
Epochal Tsukuba |
[Poster Presentation]
Task Selection for Bandit-Based Task Assignment in Heterogeneous Crowdsourcing Hao Zhang (Tokyo Tech), Masashi Sugiyama (UTokyo) IBISML2015-65 |
[more] |
IBISML2015-65 pp.91-98 |