Using the mathematics of probability to optimize machine learning and predictions.
Information and Communication Technology (Saitama University Research Seed Collection 2025-27 p.88)
Keywords: Probabilistic information processing, machine learning, data analysis, time series data, probabilistic simulation, annealing-type computer.
Artificial intelligence, machine learning, and simulation technologies are greatly changing our lives. On the other hand, there are issues such as environmental burdens from massive electricity consumption and the large-scale nature of training data. In order to significantly improve the efficiency of these information processing methods, I am conducting research that broadly explores various fields of mathematics, identifies usable technologies, and refines them into forms that can be applied in engineering. Although I am still in the research phase, I am able to accelerate predictions and controls by dozens of times using mathematics known as duality and Koopman operators, and I can compress neural networks used in artificial intelligence technologies. I am also aiming for machine learning with small amounts of data utilizing knowledge about the subject. Additionally, I am involved in research related to quantum computers known as annealing types, as well as simulations and estimations of probabilistic phenomena. Finding and refining usable mathematics can be challenging, but I aim to develop foundational technologies based on mathematics that are unique to universities.
Inquire About This Product
basic information
Jun Okubo, Professor Graduate School of Science and Technology, Department of Mathematical and Electronic Information, Information Area 【Recent Research Themes】 ● Acceleration using the duality of prediction and control for stochastic systems ● Equation estimation from time series data using Koopman operators ● Development of machine learning techniques utilizing both equations and data ● Theoretical research on annealing-type hardware
Price range
Delivery Time
Applications/Examples of results
【Appeal Points to the Industry】 ● We can collaboratively explore the application of new information processing technologies based on mathematics. ● It can be utilized for speeding up simulations of probabilistic phenomena and for parameter estimation. ● We are developing equation estimation methods from time series data. 【Examples of Practical Application, Use Cases, and Utilization】 ● (Commercialized through joint research) Development of an automatic phase analysis method for spectral analysis devices. ● (Use case) Analysis and future forecasting of time series data.
Detailed information
-
Optimization of machine learning
catalog（1）
Download All Catalogs
Company information
The Saitama University Open Innovation Center is a center that functions as a liaison office for industry-academia-government collaboration. It consists of three departments: the Industry-Academia-Government Collaboration Department, the Intellectual Property Department, and the Startup Support Department, each staffed with coordinators well-versed in various fields. The center's activities include solving technical challenges in companies, supporting the implementation of joint research, and conducting technology transfer aimed at introducing and utilizing Saitama University's intellectual property.
![[Leveling Correction] Leveling correction of concrete floors "Teratec Method"](https://image.mono.ipros.com/public/product/image/d08/1267539002/IPROS70063422645318547243.jpeg?w=280&h=280)
