レッパネン ユホ 特任講師
 |
所属東海大学 総合科学技術研究所 連絡先E-mail：leppanen[at]tsc.u-tokai.ac.jp |
資格/免許
Ph.D.
略歴
- 2015－2018
- Ph.D. Candidate,
Department of Mathematics and Statistics,
University of Helsinki (Finland) - 2018－2020
- Postdoctoral Researcher,
Laboratory of Probability, Statistics, and Modelling, Sorbonne University (France) - 2020－2021
- Data Scientist, XICA inc. (Japan)
- 2021－2022
- Data Scientist, Welmo inc. (Japan)
- 2022－Present
- 東海大学 総合科学技術研究所 特任講師
Junior Associate Professor,
Research Institute of Science and Technology, Tokai University (Japan)
研究分野
I use methods from probability, stochastics, and functional analysis to describe the statistical properties of chaotic systems, i.e. systems that display a sensitive dependence on initial conditions. The term statistical refers to the long-term behavior of trajectories of such systems, e.g., the limiting behavior of time series arising as observations along the trajectory, frequencies of deviations from these limits, etc.
Most of my research so far has dealt with probabilistic limit laws, such as central limit theorems and concentration inequalities, associated to partial sums of (piecewise) smooth uniformly and non-uniformly hyperbolic systems, with particular focus on non-autonomous and intermittent type dynamics. I have also conducted research related to response theory, which concerns the robustness of statistical properties of dynamical systems subject to perturbations. My ongoing research focuses on obtaining quantitative extensions of certain limit laws for hyperbolic dynamical systems. Such results give concrete estimates on the distance between the underlying process and the limiting distribution.
キーワード
双曲線力学、統計限界法、断続的力学、非自律力学、応答理論、転送演算子、マーチンゲール
Hyperbolic dynamics, Statistical limit laws, Intermittency,Non-autonomous dynamics, Response theory, Transfer operator
所属学会
日本数学会
The Mathematical Society of Japan
メッセージ
Dynamical systems theory has found applications in almost all areas of science and technology. A modern example can be found in machine learning, where dynamical systems theory has been useful in improving the interpretability and efficiency of certain algorithms, such as recurrent neural networks used to model sequential data. The machine learning approach becomes important in the case of complex dynamics such as those displayed by the climate or brain, in which case explicit models are hard to construct. I would be highly interested in contributing to such interdisciplinary studies in the future.
