Joint Research Projects
- The Bank of Tokyo-Mitsubishi UFJ, Ltd.
- Nomura Securities Co., Ltd.
- The University of Tokyo
AI (artificial intelligence) is the science and engineering of making intelligent machines.
Data mining is the process of discovering interesting patterns or knowledge from data.
Machine learning is a field of AI research focusing on getting computers (machines) to learn.
Bioinformatics is a scientific discipline focusing on biological information.
Reinforcement learning is a method to learn actions based on trial and error.
Our research topics on reinforcement learning include:
- Compound reinforcement learning that is a new framework of reinforcement learning to maximize the compound returns,
- Learning trage strategies in financial markets using reinforcement learning, and
- OnPS (on-line profit-sharing) algorithm that updates parameters in each step and can be implemented with a bounded memory size, but is equivalent to ordinary (off-line) profit-sharing.
Text mining is a subfield of data mining focusing on text data.
Our research topics on text mining include:
- Analyzing Band of Japan monthly reports to predict the future interest using coocasion, principal component analysis (PCA), and regression,
- Text categorization for Nikkei newspapers to predict the future markets,
- Information retrieval to find users' reviews from weblogs.
Financial Data Mining
Financial data mining is a subfield of data mining focusing on financial data.
Our research topics on financial data mining include:
- Analyzing trage strategies in financial markets and technical indices used in the strategies using reinforcement learning,
- Predicting the future stock market by categorizing Nikkei newspapers, and
- Predicting the future interest by analyzing Bank of Japan monthly reports using coocation, principal component analysis (PCA), and regression.
Bio Data Mining
Bio data mining is a subfield of data mining focusing on biological data or data mining in bioinformatics.
Our research topics on bio data mining include:
- Homology detection tool for proteins with multiple functional domains,
- Categorizing proteins into folds from the primary structures using inductive logic programming (ILP),
- Recommending similar mutants using association rule mining from purchase records.