This course offers an intensive exploration of AI and Machine Learning (ML), focusing on their transformative roles across
various domains. Building on an introduction to AI and ML principles, the course delves into advanced applications, showcasing
how these technologies are revolutionizing fields such as healthcare, autonomous vehicles, and security systems. The rest
of the curriculum is dedicated to addressing the critical safety and ethical challenges posed by AI deployment. Through discussions
and case studies students will examine the potential risks, biases, and ethical dilemmas inherent in AI systems, learning
strategies to mitigate these issues and ensure the responsible use of AI technology.
In this course, students are divided into groups to:
(1) As home work: Read the papers, discuss with your team, make presentations
(2) at the class: present and discuss
Note that:
- This class is quarter-based whose duration is half of a semester (7 weeks). Each week will cover the content which is equivalent to those of two schedule classes. (will long from 9:00 - 12:30).
- In this class, several researchers, professors from both industry and academia might also join as guest lecturer (e.g., from Google, Apart Research, Tokyo AI Safety, etc.).
In this course, students are divided into groups to:
(1) As home work: Read the papers, discuss with your team, make presentations
(2) at the class: present and discuss
Note that:
- This class is quarter-based whose duration is half of a semester (7 weeks). Each week will cover the content which is equivalent to those of two schedule classes. (will long from 9:00 - 12:30).
- In this class, several researchers, professors from both industry and academia might also join as guest lecturer (e.g., from Google, Apart Research, Tokyo AI Safety, etc.).
The purpose of class is to provide students with a well-rounded understanding of both the opportunities and challenges presented
by AI technologies. Additionally, this class aim to equip students paper-reading skills, presentation skills, etc.
- The students can understand the basics of AI and Machine Learning
- The students can understand the wide range applications of AI
- The students can build up critical thinking about the ethical considerations and safety issues surrounding the deployment of artificial intelligence systems
| Presentation | Discussion | Total. | |
|---|---|---|---|
| 1. | 20% | 20% | 40% |
| 2. | 15% | 15% | 30% |
| 3. | 15% | 15% | 30% |
| Total. | 50% | 50% | - |
| Class schedule | HW assignments (Including preparation and review of the class.) | Amount of Time Required | |
|---|---|---|---|
| 1． | Introduction to Machine Learning (I) | Investigate the basic concept of ML including neural networks, gradient descent, forward and backward propagation. | 190minutes |
| 2． | Introduction to Machine Learning (II) | Investigate the basic concept of ML including neural networks, gradient descent, forward and backward propagation. | 190minutes |
| 3． | Fundamental models and their applications (I) | Investigate the basic models including CNN, LSTM and their roles in practical applications | 190minutes |
| 4． | Fundamental models and their applications (II) | Investigate the basic models including CNN, LSTM and their roles in practical applications | 190minutes |
| 5． | Advanced models and their applications (I) | Investigate the recent advanced models: transformer, diffusion. Everything LLMs: applications, variations, what can go wrong, how do we know and what to do about it ? |
190minutes |
| 6． | Advanced models and their applications (II) | Investigate the recent advanced models: transformer, diffusion. Everything LLMs: applications, variations, what can go wrong, how do we know and what to do about it ? |
190minutes |
| 7． | The AI Alignment problem - an overview (I) | Investigate issues of AI safety and how to tackle the alignment problems. | 190minutes |
| 8． | The AI Alignment problem - an overview (II) | Investigate issues of AI safety and how to tackle the alignment problems. | 190minutes |
| 9． | Reinforcement learning from human feedback | Investigate the basic concepts of Artificial General Intelligence (AGI) and the AI Alignment problem - Outer & inner alignment problems: reward misspecification & instrumental convergence, deception, mesa-optimizer, goal misgeneralization etc. - Technical alignment: RLHF, Constitutional AI, debate, etc. |
190minutes |
| 10． | Scalable Oversight | Investigate the basic concepts of Artificial General Intelligence (AGI) and the AI Alignment problem - Outer & inner alignment problems: reward misspecification & instrumental convergence, deception, mesa-optimizer, goal misgeneralization etc. - Technical alignment: RLHF, Constitutional AI, debate, etc. |
190minutes |
| 11． | Mechanistic Interpretability | Investigate the basic concepts of Artificial General Intelligence (AGI) and the AI Alignment problem - Outer & inner alignment problems: reward misspecification & instrumental convergence, deception, mesa-optimizer, goal misgeneralization etc. - Technical alignment: RLHF, Constitutional AI, debate, etc. |
190minutes |
| 12． | AI Governance | Investigate the need for technical AI governance work. | 190minutes |
| 13． | Final Presentation (I) | Summarize what you have learnt, how you can use the acquired knowledge for your future, and so on. Working with your group to write a paper-like report and prepare the final presentation for your group. | 190minutes |
| 14． | Final Presentation (II) | Summarize what you have learnt, how you can use the acquired knowledge for your future, and so on. Working with your group to write a paper-like report and prepare the final presentation for your group. | 190minutes |
| Total. | - | - | 2660minutes |
<Grading Criteria>:
- Presentation and communicative ability: 50%
- Discussion: 50% (where in-class discussion is 30%, in-group discussion is 20%).
- A total score of more than 60% is required to earn course credit.
<Note>:
- Students will be marked absent if they arrive more than 5 minutes late to class.
- If students arrive late (by less than 5 minutes) twice, it will be counted as one absence
- If a student is absent for more than one-third of the total number of classes, they will not be eligible to receive course credit.
- Any absence will be counted as such unless valid evidence (e.g., a doctor’s note for illness or a similar justified reason) is provided.
- Presentation and communicative ability: 50%
- Discussion: 50% (where in-class discussion is 30%, in-group discussion is 20%).
- A total score of more than 60% is required to earn course credit.
<Note>:
- Students will be marked absent if they arrive more than 5 minutes late to class.
- If students arrive late (by less than 5 minutes) twice, it will be counted as one absence
- If a student is absent for more than one-third of the total number of classes, they will not be eligible to receive course credit.
- Any absence will be counted as such unless valid evidence (e.g., a doctor’s note for illness or a similar justified reason) is provided.
| ways of feedback | specific contents about "Other" |
|---|---|
| Feedback in the class |
Machine Learning Specialization (Coursera) - Stanford University
https://www.coursera.org/specializations/machine-learning-introduction#courses
https://www.coursera.org/specializations/machine-learning-introduction#courses
- Contact based on the appointments via Prof. Phan Xuan Tan's email: tanpx@shibaura-it.ac.jp
| Work experience | Work experience and relevance to the course content if applicable |
|---|---|
| N/A | N/A |
Last modified : Tue Mar 11 04:09:09 JST 2025