招待講演

招待講演1：Data analytics in industrial research: a personal retrospective

概要: In this talk I will present a personal retrospective on the role that data analytics and machine learning have played in industrial research, based on my experience on both sides of the Pacific. Starting with some earlier work I was involved in, while with NEC Research in the 90’s, and moving onto projects I worked on at IBM Research in 2000’s, I will touch upon a number of sub-topics of machine learning and associated application areas, reflect on their success, failure and lessons learned. I will also discuss some of our on-going activities and trends in emerging areas, such as open data analytics and smarter agriculture. Through these reflections, I hope to shed some light on what constitutes a workable model of industrial research in the area of induction sciences.

招待講演2：Modeling-based dataset retrieval

概要: A challenge of data-driven sciences is how to make maximal use of growing databases of experimental datasets to keep research cumulative. We introduce modeling-based methods for retrieving data sets relevant to a query dataset, usable in particular for relating an empirical researcher’s experimental dataset to earlier work in the field. The search is (i) data-driven to enable new findings, going beyond keyword searches in annotations, and (ii) modeling-driven, to take into account both prior knowledge and insights learned from data. I will discuss case studies on gene expression data sets, and alternative formulations of the underlying probabilistic modeling and inference problem.

招待講演3：理論研究とアルゴリズム・機械学習・AI

概要: 近年、アルゴリズム・理論計算機科学の研究者が機械学習・AI分野に参入し、多くの成果を挙げている。その理由として主に二つが挙げられる。

  • (1) 「最適化」問題をセンスよく、かつ効率的に解く技術を多数もっている。
  • (2) データサイエンスから派生する問題を、（近似的に）解決可能に「定式化」 する技術を多数持っている。

私が研究総括を務めるERATO「河原林巨大グラフプロジェクト」でも、２０１３年より、若手理論研究者が、機械学習、AI,　データーマイニング、 データーベースなどの分野で、世界的研究成果を挙げようと試みてい る。
本講演では、過去２年半の「試み」で成功した点、改善点などを報告したい。