Keynote
Presenter:
Prof. Gen Chiaki
National Institute of Technology, Kochi College (Kochi Kosen)
Title:
Metal enrichment and star formation in the early Universe
Abstract:
Completely different from the local Universe, just after the Big Bang, there were no light sources, such as stars or galaxies, and the Universe was dark. Also interesting, no elements heavier than helium
(called “metals” in the field of astrophysics) had been created as predicted by the Big Bang nucleosynthesis theory. The first astronomical object is a star, called the first star, which formed in
gravitationally collapsing gas clouds typically 0.1 Gyr after the Big Bang. It plays a crucial role in emitting the first photons and synthesizing the first metals. If the star is massive (more than 8 times the solar mass), it ends its life with an explosive phenomenon (supernova explosion). Then, the metals synthesized in the stellar mantle can be released into the interstellar medium. This is the very first step of metal enrichment until the metal fraction (metallicity) eventually reaches 2% as in the local Universe.
How massive was the first star and how were the metals dispersed? These are very important questions to know the origin of metals from which our planet and life are made. I have studied the star formation
and metal enrichment processes in the very early Universe (0.1-1 Gyr after the Big Bang) making use of the latest numerical simulation techniques. First, I found that the masses of embryo stars become
massive in environments with lower metal fractions. This suggests that the typical mass of the first stars are massive (10-100 times the solar mass). Also, the metals dispersed from the first supernova
mainly enriches the cloud that has hosted the star. The enriched cloud collapses again, and forms the second generation of stars with slight fractions of metals (less than 1/1000 times solar metallicity).
Biography:
Gen Chiaki is working on star and galaxy formation mainly in the first billion years after the Big Bang by making use of the latest numerical simulation techniques. He got his PhD in the University of Tokyo in 2016 supervised by Naoki Yoshida. In his first postdoc, he worked in Konan University in Kobe with Hajime Susa. Then, he moved to Georgia Institute of Technology, US, and studied for three years and half with his mentor, John H. Wise. He came back to Japan and belonged to Tohoku University and National Astronomical Observatory of Japan for one year each. Currently he is an assistant professor in National Institute of Technology, Kochi College (Kochi Kosen).
Invited Talk
Presenter:
Prof. Soemsak Yooyen
Department of Aeronautical Engineering,
International Academy of Aviation Industry,
King Mongkut’s Institute of Technology Ladkrabang (KMITL)
Title:
Development of a neural-network-based and a statistic model for estimating airplane contingency fuel
Abstract:
Airplane contingency fuel is the amount of fuel used to compensate for unexpected events. The estimation of airplane contingency fuel in the form of the lowest and the highest contingency fuel has been determined without statistical reference. A statistical model uses statistical inference to calculate the contingency fuel in the form of Statistical Contingency Fuel (SCF). This is a method of calculating the amount of fuel using a statistical method that covers the deviation from the planned to actual trip fuel. This is calculated from the historical statistics at various confidence levels and use loss functions to evaluate. This research also develops a neural network to predict fuel burn, which is learned from historical airplane data. The experiment applies to local and international flight data. The predicted model is swapped and tested on the outbound and inbound replacements for confirmation capable of the predicted mode. This research will help prevent wasted fuel burning. In addition, it can help saving fuel and cost by preparing the lowest amount of contingency fuel which follow the policy of Civil Aviation Authority of Thailand (CAAT).
Biography:
Dr. Soemsak Yooyen obtained his PhD at Mie University, Department of Mechanical Engineering, Graduate Program in Quantum and Nano Science in 2013. He is currently a professor (Assistant) of the Department of Aeronautical Engineering of King Mongkut’s Institute of Technology Latkrabang and the Dean of International Academy of Aviation Industry. His research interests cover many areas of quantum plasma, new engineering materials synthesis, aeronautical and aerospace engineering as well as innovation in aviation application.
Invited Talk
Presenter:
Prof. Shutaro Takeda
Urban Institute, Kyushu University
Title:
How can machine learning contribute to solving real-life social issues?
Abstract:
Our international joint research between Kyushu University, Waseda University, and Harvard University is conducting interdisciplinary research aiming to quantify human rights by applying machine learning to econometrics.
The International Labour Organization (ILO) estimates that about 25 million people are trapped in forced labor worldwide, while 152 million children are victims of child labor. However, scientific methods to quantify human rights violation risks have not been established (Barsh, 1993; Landman, 2004), and thus it has been difficult to incorporate human rights into econometrics (Fariss & Dancy, 2017).
Particularly challenging is quantifying the extent to which risks of human rights violations, ranging from child labor to gender equality, occur in which countries and in which industry sectors. Social Hotspot Database (Benoit-Norris et al., 2012), based on guidelines published by the United Nations Environment Programme in 2009, has several research and practical applications, but its risks are assessed by researchers manually reading the literature of over 400 international organizations.
We are therefore working to quantify human rights violation risks by analyzing hundreds of thousands of daily published news with LLM. This is accomplished with our patent-pending CPT (Classification Pre-trained Transformers) algorithm, a large language model with 128 million parameters that analyze 30 human rights-related topics from news reports. Our CPT model shows approximately 95% cross-validation accuracy, and can analyze 10,032 news items published in the U.S. daily in real-time.
With our machine learning model, international organization, governments and NPOs can not only identify the global hotspots of human rights but also quantify the labor hours at risk in a specific country in a specific industry sector in real time.
Biography:
Prof. Shutaro Takeda is an Associate Professor at the Urban Institute, Kyushu University. Formerly a United Nations officer and an international award-winning young scientist, Prof. Takeda has been working to establish a new science named “Sustainametrics,” after finishing his master’s at Harvard and Ph.D. at Kyoto University.
Sustainametrics is an approach that merges data science with sustainability efforts to address global challenges and achieve sustainable development goals (SDGs). It involves the use of advanced data analysis techniques, including artificial intelligence and machine learning, to evaluate and monitor environmental, social, and economic aspects of sustainability more comprehensively and quickly than was possible before.