Future Wisdom Institute - Informatics-based interdisciplinary research group

Template of a research paper (including abstract)

Notice A student shall demonstrate the ability to do original research by himself/herself and publish at least three papers in five years (at most six years). Usually in the first three years one shall follow the instruction of Prof. Zhao. For who is not sure what topic to work with, please consider to start from a project in this page (including those pending ones). For a quick glance, see research.pdf.

Wisdom and wise AI

What is wisdom and what is its difference from knowledge? Zhao proposes a theory Wisdom = Learning + Random selecting (thus, roughly speaking, the difference is the freedom of doing random choice). To study it, we are interested in how boids, human, autonomous devices, creativity and future wise lifes can learn rules for cooperation. Latest publication: Zhao L., Li W. (2020) “Choose for No Choose”—Random-Selecting Option for the Trolley Problem in Autonomous Driving. In: Zhang J., Dresner M., Zhang R., Hua G., Shang X. (eds) LISS2019. Springer, Singapore. https://doi.org/10.1007/978-981-15-5682-1_48 (a draft version is available at researchgate).

Keywords: Wisdom,AI, Algorithm, Philosophy, Cognition, Creation, Random

Distance domination and representative nodes in complex networks

Suppose Facebook needs to have a users' meeting. We must first decide how many representatives shall be invited. The more the better democracy but the less efficient. So what is a convenient number? The problem has actually been studied for a long history about the representativers of the public. We developed a simple model using social network and are working on its applications (well, more than just applilcations). The latest publication: Zhao & Peng, An Allometric Scaling for the Number of Representative Nodes in Social Networks (a draft version is available on researchgate).

Parliament size and malapportionment

What is a convenient size of the parliament? How to distribute the seats to each districts fairly? These are the issues we are studying. Motivated by our research on representative nodes in networks, we have proposed some models to study these issues. The latest publication: (under preparation).

Keywords: Parliament size, Representative, Malappointment, 議員定数, 代表数, 一票の格差

Graph learning

We study methods and applications of graph Learning, i.e., learning features (finding orders) from graph instances automatically with modern machine/deep learning techniques. One application is algorithms for chemical compound inferring and searching (QSAR and inverse QSAR) by collaboration with Nagamochi lab; Another application is algorithms for drug discovery from Traditional Chinese Medicine.

Image processing and machine learning algorithm for screening of diseases

We are developing image processing and machine learning algorithm for screening of diseases such as mental disorder (see the detail) and glaucoma.

Keywords: Image processing, Machine learning, Mental disorder screening, Glaucoma screening

Projects wanting participants

Optimization in evacuation shelter assignment/避難所割当最適化


Keywords/キーワード: 避難所割当,最適化,自主防災

Level: undergraduate or master project (3 months or more)

Approximation algorithm and its implementation for the Minimum Equivalent Graph problem

Minimum Equivalent Graph problem asks to find a minimum-size, spanning subgraph of a directed graph that keeps the reachability among pairs of nodes. This problem has applications in network design, network visualization, etc. Prof. Zhao developed a very fast algorithm for this NP-hard problem long time ago and would like to provide an open-source implementation to, e.g., the NetworkX package.

Keywords: minimum equivalent graph, strongly-connected spanning subgraph, union-find structure

Level: internship project for an undergraduate or master student (1 month)

Past projects

  • Easy screening for secondary hypertension (collaboration with 新疆人民医院. 日本笹川医学奨学金第40期共同研究コース)
  • Early screening method for dementia