2026 (Current Year) Special graduate degree programs Specially Offered Degree Programs for Graduate Students Center of Data Science and Artificial Intelligence

Progressive Applied Practical Data Science and Artificial Intelligence 3C

Academic unit or major: Center of Data Science and Artificial Intelligence
Instructor(s): Katsumi Nitta / Takayoshi Yokota / Kei Miyazaki / Keiji Okumura / Yutaro Tachibana / Yoshihiro Miyake / Jun Sakuma / Isao Ono / Hirotaka Uchitomi / Norio Tomii
Class Format: Lecture (HyFlex)
Media-enhanced courses: -
Day of week/Period (Classrooms): 7-8 Fri (M-B07, G2-202)
Class: -
Course Code: DSA.P633
Number of credits: 100
Course offered: 2026
Offered quarter: 3Q
Syllabus updated: Sep 17, 2026
Language: Japanese

Syllabus

Course overview and goals

This course aims to provide students with an understanding of the current state of social implementation and the latest developments in artificial intelligence and data science technologies, and to encourage them to consider the potential applications and challenges of these technologies. In each class, lecturers from Institute of Science Tokyo will introduce case studies of technology and product development utilizing data science and artificial intelligence.

Through acquiring knowledge of applications such as large language models, Physical AI, and train timetable optimization, and explaining their own considerations regarding social applications in course reports, students are expected to develop a broad perspective that will enable them to play an active role in the real world.

This course places particular emphasis on dialogue with the lecturers. In addition to the seven regular classes, students are expected, in principle, to participate in the DS&AI Forum, which will be held at the Ookayama Campus in late November.

Course description and aims

This course aims to develop ability of each student to be more successful in the real world with the consideration of social implementation of data science and artificial intelligence.

Student learning outcomes

実務経験と講義内容との関連 (又は実践的教育内容)

This course is mainly taught by specially appointed faculty members of Science Tokyo, based on practical experience.

Keywords

Data Science, Artificial Intelligence, Large Language Model, robotics, optimization

Competencies

Specialist skills
Intercultural skills
Communication skills
Critical thinking skills
Practical and/or problem-solving skills

Class flow

This course is classified as a high-flex type, but can only be taken in designated classrooms in Ookayama and Suzukakedai.

Course schedule/Objectives

	Course schedule	Objectives
Class 1	Understanding and Building LLMs with nanoGPT: A Hands-on Approach	The fundamental mechanisms of Large Language Models (LLMs) are explained, focusing on next-token prediction and sentence generation. Using the open-source software nanoGPT, students create and pretrain a small language model on a favorite novel. Through this hands-on experience, students learn the concepts of corpora and tokenizers, and how LLMs acquire text generation ability.
Class 2	Understanding Transformers through a Visual Walk-Through	By observing the processing steps in nanoGPT, students learn the basic mechanisms of the Transformer, including token embeddings, positional embeddings, and the attention mechanism. Through hands-on exercises, students learn how input tokens are processed by the Transformer and how the next token is predicted.
Class 3	Understanding Instruction Following through Instruction Tuning	Using an LLM pretrained on a larger corpus, students learn about Supervised Fine-Tuning (SFT), which teaches the LLM how to respond to questions and instructions. Through hands-on exercises, students learn how SFT transforms a pretrained LLM from a simple text generator into a model that can respond appropriately to users' questions and instructions.
Class 4	Introduction to Physical AI 1	Can Embodied Intelligence Surpass LLMs? ? Rethinking the Relationship Between Intelligence and the Body Develop a perspective that understands intelligence not as symbolic processing within the brain, but as something that emerges through interactions between the body and its environment.
Class 5	Introduction to Physical AI 2	TThe Frontiers of Human Motion Measurement ? AI for High-Precision Estimation of Gait and Posture As a first step toward understanding embodied intelligence, learn the technologies for quantitatively measuring and estimating human movement, as well as the deep learning models that underpin them.
Class 6	Introduction to Physical AI 3	The Frontiers of Physical Intervention ? Co-Creative Walking Assist Robots Toward applying embodied intelligence, learn the design principles of co-creative interactions in which humans and robots collaborate to generate movement.
Class 7	Application of Optimization Techniques to Real-World Problems: Railway Rolling Stock Scheduling	Real-world problems are often complex. To address such problems, it is necessary not only to acquire domain knowledge but also to consider how the methods presented in textbooks can be applied to individual problems. In this lecture, using railway rolling stock scheduling as an example, we will learn how optimization algorithms can be applied to address complex real-world problems. Please read the materials provided in advance. Also, please make sure that you have an environment in which you can use Google Colaborator

Study advice (preparation and review)

To enhance effective learning, students are encouraged to spend approximately 100 minutes preparing for class and another 100 minutes reviewing class content afterwards (including assignments) for each class.

Textbook(s)

None required.

Reference books, course materials, etc.

Materials will be provided on Science Tokyo LMS in advance.

Evaluation methods and criteria

No final exam will be given. Grades will be evaluated based on each assignment report, the term-end report and the participation report of the DS&AI Forum scheduled for late November. Please note that it is not possible to submit assignment reports for missed lectures. Even if a student submits an assignment report for a lecture he/she has missed, it will not be graded.

Related courses

XCO.T677 ： Fundamentals of Progressive Data Science
XCO.T678 ： Exercises in Fundamentals of Progressive Data Science
XCO.T679 ： Fundamentals of Progressive Artificial Intelligence
XCO.T680 ： Exercises in Fundamentals of Progressive Artificial Intelligence

Prerequisites

As this course includes practical exercises, the following prerequisites are assumed for students.
・Have basic IT literacy.
・Be able to use Python, Jupyter Notebook, and related tools.

Only students of doctor curse are acceptable. Other students must take DSA.P433 " Applied Practical Data Science and AI 3C" instead of this course.

Contact information (e-mail and phone) Notice : Please replace from ”[at]” to ”@”(half-width character).

Katsumi Nitta, Takayoshi Yokota
lecture_ap[at]dsai.isct.ac.jp

Office hours

Contact by e-mail in advance to schedule an appointment.

Other

・This class is a technical course that can be considered an entrepreneurship course. The GAs that this subject corresponds to are GA0D and GA1D.