Ebihara Lab Homepage

Study Topics

We explore analysis and synthesis of dynamical systems from broad perspective using mathematical systems theory. Currently, we are actively working on the reliability/stability verification of AI systems on the basis of optimization and control theory.

Special Events

News

September 2024: Journal Paper Accepted New!!
The following paper that deals with the verification of NN-driven control systems have been accepted for publication at Transactions of the Society of Instrument and Control Engineers,
  • T. Yuno, K. Fukuchi, and Y. Ebihara,
    Construction and Input-to-Output Characteristics Evaluation of Compressed Model of Recurrent Neural Networks for their Stability Analysis (in Japanese)
July 2024: Conference Papers Accepted New!!
The following papers that deal with the verification of NN-driven control systems have been accepted for presentation at IEEE Conference on Decision and Control 2024.
  • T. Yuno, K. Fukuchi, and Y. Ebihara,
    A Lyapunov-based Method of Reducing Activation Functions of Recurrent Neural Networks for Stability Analysis
  • T. Yuno, S. Nishinaka, R. Saeki, and Y. Ebihara,
    On Static O'Shea-Zames-Falb Multipliers for Idempotent Nonlinearities
June 2024: Conference Paper Accepted
The following paper that deals with the reliability verification of neural networks has been accepted for presentation at The 4th IFAC Conference of Modeling, Identification and Control of Nonlinear Systems (MICNON2024).
  • T. Yuno, S. Nishinaka, R. Saeki, Y. Ebihara, V. Magron, D. Peaucelle, S. Zoboli, and S. Tarbouriech,
    Stability Analysis of Feedback Systems with ReLU Nonlinearities via Semialgebraic Set Representation
April 2021: New Research Project Launched
"Stability Analysis and Optimal Synthesis of Recurrent Neural Networks by Conic Programming"
Collaborators: H. Waki (IMI, Kyushu University, Japan), D. Peaucelle, S. Tarbouriech, J. B. Lasserre, V. Magron, N. Mai (LAAS-CNRS, France).