OMA Syllabus

Virtual Meeting Schedule: Every other week, starting January 14, 2025, 7:00 - 8:00 PM (Eastern) via Zoom


Class Dates:

Start: January 6, 2025
Spring Break: March 17 - March 21, 2025
End: April 22, 2025
Course Description:


This advanced graduate seminar explores the theoretical foundations, computational frameworks, and emerging paradigms of multi-agent systems (MAS) and collaborative artificial intelligence. Students will critically examine how autonomous agents interact, negotiate, coordinate, and collectively solve complex problems across diverse domains.

Course Goals

By the end of this course, students will:

Gain an in-depth understanding of multi-agent systems (MAS) and collaborative intelligence.
Explore key concepts such as agent design, communication protocols, coordination mechanisms, and ethical implications in MAS.
Apply theoretical knowledge to practical problems in various domains where multi-agent systems can provide solutions.
Develop the ability to critically analyze and evaluate the impact of MAS on social, economic, and ethical dimensions.
Enhance communication and collaboration skills by engaging in class discussions and interacting with peers.
Learning Outcomes

By the end of this course, students will be able to:

Define and explain the core concepts of multi-agent systems
Describe the components and architecture of a multi-agent system, including reactive and deliberative agents.
Explain the principles of collaborative intelligence in multi-agent environments.
Analyze different coordination mechanisms and their applications
Evaluate various coordination techniques, including auctions, negotiation, and market-based mechanisms.
Identify the advantages and disadvantages of each mechanism in different scenarios.
Design and simulate multi-agent systems
Develop a basic framework for agent design, including behaviors, communication, and decision-making strategies.
Implement simple coordination algorithms and agent interactions in simulation environments.
Critically assess the ethical implications of MAS
Discuss the ethical considerations in the design of multi-agent systems, including fairness, transparency, and trust.
Identify potential risks related to privacy, security, and societal impacts in the deployment of MAS.
Apply multi-agent systems to real-world problems
Apply MAS to solve real-world challenges, such as autonomous vehicle coordination, resource allocation, and healthcare delivery.
Use case studies to understand the practical applications and limitations of MAS in various domains.
Demonstrate effective communication and teamwork skills
Participate actively in class discussions, providing thoughtful insights and feedback to peers.
Collaborate effectively in team-based activities and discussions about multi-agent systems and collaborative intelligence.
Weekly Topics, Discussions, and Schedule

Week 1 (No Virtual Class Meeting):

Topic: Introduction to Multi-Agent Systems
Discussion Question: What is a multi-agent system, and why is it important for collaborative intelligence? Provide examples from real-world systems.
Week 2:

Topic: Agent Design and Architectures
Discussion Question: What are the key differences between reactive and deliberative agent architectures? How do these affect the behavior and efficiency of a multi-agent system?
Virtual Class Meeting - Tuesdays - 7:00 - 8:00 PM Eastern (Check class announcements)
Week 3 (No Virtual Class Meeting):

Topic: Coordination Mechanisms in Multi-Agent Systems
Discussion Question: Compare and contrast different coordination mechanisms (e.g., auctions, negotiation, and market-based methods) used in multi-agent systems. Which would you recommend for an environment with scarce resources?
Week 4:

Topic: Communication in Multi-Agent Systems
Discussion Question: How does communication between agents influence the success of a multi-agent system? Discuss the importance of communication protocols in ensuring system reliability and efficiency.
Virtual Class Meeting - Tuesdays - 7:00 - 8:00 PM Eastern (Check class announcements)
Assignment 1: Basic Agent Framework (Optional)
Week 5 (No Virtual Class Meeting):

Topic: Conflict Resolution in Multi-Agent Systems
Discussion Question: In multi-agent systems, how can agents resolve conflicts without human intervention? What strategies would work best in high-stakes, competitive environments?
Week 6:

Topic: Learning and Adaptation in Multi-Agent Systems
Discussion Question: How can reinforcement learning be applied to multi-agent systems? Discuss the challenges of implementing learning in collaborative vs. competitive environments.
Virtual Class Meeting - Tuesdays - 7:00 - 8:00 PM Eastern (Check class announcements)
Week 7 (No Virtual Class Meeting):

Topic: Game Theory in Multi-Agent Systems
Discussion Question: How can game theory be applied in multi-agent systems for decision-making? Discuss examples where game theory has helped optimize agent behavior.
Week 8:

Topic: Ethical Considerations and Trust in Multi-Agent Systems
Discussion Question: How does trust play a role in the interactions between agents? Can agents "build" trust in the same way humans do, and what ethical challenges arise when designing MAS?
Virtual Class Meeting - Tuesdays - 7:00 - 8:00 PM Eastern (Check class announcements)
Assignment 2: Coordination Mechanism Implementation (Optional)
Week 9 (No Virtual Class Meeting):

Topic: Applications of Multi-Agent Systems in Autonomous Vehicles
Discussion Question: How do multi-agent systems facilitate the coordination of autonomous vehicles? What are the key challenges and benefits of using MAS in transportation?
Week 10:

Topic: Collective Intelligence and Collaborative Agents
Discussion Question: Discuss how collective intelligence emerges from the interactions of individual agents. How can MAS be used to improve decision-making in large-scale systems?
Virtual Class Meeting - Tuesdays - 7:00 - 8:00 PM Eastern (Check class announcements)
Week 11: Spring Break

Week 12:

Topic: Social and Economic Implications of MAS
Discussion Question: How do multi-agent systems impact economies and societies? Consider implications such as job displacement, the role of automation, and decision-making biases.
Virtual Class Meeting - Tuesdays - 7:00 - 8:00 PM Eastern (Check class announcements)
Week 13 (No Virtual Class Meeting):

Topic: MAS for Health and Social Good
Discussion Question: How can multi-agent systems be applied to solve pressing social issues like healthcare delivery, disaster response, or homelessness? Discuss potential benefits and challenges.
Week 14:

Topic: Privacy, Security, and Multi-Agent Systems
Discussion Question: What are the security concerns when designing multi-agent systems? How can you ensure privacy and data protection in collaborative agent environments?
Virtual Class Meeting - Tuesdays - 7:00 - 8:00 PM Eastern (Check class announcements)
Assignment 3: Reinforcement Learning for Cooperation (Optional)
Week 15 (No Virtual Class Meeting):

Topic: Future Directions and Challenges in Multi-Agent Systems
Discussion Question: What are the key challenges facing the future of multi-agent systems? Consider advancements in technology, ethical concerns, and the role of human oversight in MAS.
Week 16:

Topic: The Future of Collaborative Intelligence
Discussion Question: How do you see collaborative intelligence evolving over the next 5-10 years? What role will AI play in the development of these systems?
Virtual Class Meeting - Tuesdays - 7:00 - 8:00 PM Eastern (Check class announcements)
Optional Coding Assignments (Complete on Your Local Machine)

Assignment 1: Basic Agent Framework (Optional)

Objective: Implement a simple multi-agent system using Python or a similar language, where agents can interact based on predefined rules.

Description: Create a framework where agents can perform basic tasks such as moving around a grid and interacting with other agents.

Performance Objectives:

Understand how to design basic agent behavior and interactions.

Implement basic communication protocols between agents.

Assignment 2: Coordination Mechanism Implementation (Optional)

Objective: Implement a simple coordination mechanism, such as a market-based auction, where agents can bid for resources.

Description: Develop a system in which agents can offer bids for shared resources and select the winning bids based on predefined rules.

Performance Objectives:

Develop a coordination system and understand how agents can work together to allocate resources.

Evaluate the fairness and efficiency of the coordination mechanism.

Assignment 3: Reinforcement Learning for Cooperation (Optional)

Objective: Implement a basic reinforcement learning algorithm where agents learn to cooperate for a shared goal.

Description: Create a multi-agent environment where agents use reinforcement learning to maximize their collective reward.

Performance Objectives:

Understand the challenges of using RL in cooperative environments.

Implement and test a basic reinforcement learning model for cooperative behavior.

Grading and Participation

Discussion Participation:

Each week, students will respond to the discussion question.
Weekly Participation Requirements:
Post 1: A thoughtful and comprehensive initial post (answering the discussion question).
Post 2: A follow-up post engaging with another student’s response or expanding on the initial discussion.
Posting Rubric (100% per week):
Initial Post: 60% - Clear, well-supported, and thoughtful response.
Follow-Up Post: 40% - Engaging and meaningful response to others.
Pass: 70% or above
Students must demonstrate sufficient understanding and participation to meet course expectations. This includes posting thoughtful and meaningful responses to discussion questions, as well as engaging in discussions with peers.
Fail: Below 70%
Students who do not meet the course requirements for participation or whose contributions are insufficient will fail.
Grading Criteria for Pass/Fail:

Discussion Participation:
Regular and thoughtful responses to discussion questions.
At least two posts each week (initial and follow-up).
Posts should contribute meaningfully to the conversation, showing an understanding of the topic.
Overall Engagement:
Active participation in all class discussions.
Demonstrated critical thinking and engagement with peers' posts.
Timing Guidelines for Discussion Posts

Post Availability:
Discussion Questions Open: Each week, the discussion questions will open on Sunday at 11:59 PM (Eastern).
Discussion Questions Close: All discussion questions will close on Sunday at 11:59 PM (Eastern) of the following week.
Access to the discussion questions will be limited outside of this timeframe. Students will not be able to post after the closing time unless they have received an accommodation.
Posting Timeline:
Initial Post: Students should make their initial post by Wednesday at 11:59 PM (Eastern). This will allow enough time for others to read and engage with the post before the follow-up is required.
Follow-up Post: A follow-up response to another student’s post or the instructor’s response is due by Friday at 11:59 PM (Eastern). The follow-up should be thoughtful, contributing to the ongoing discussion.
Late Posts:
Any posts made after Sunday at 11:59 PM (Eastern) will not be accepted unless the student has previously arranged an accommodation with the instructor.
Engagement Reminder:
Students must make at least two posts per week: one initial post and one follow-up post.
Active participation is required throughout the entire discussion window to receive full credit for the week.
Course Policies

Communication:

Students are responsible for keeping up with course announcements via Zoom, email, and the course platform.
Respond to emails and discussion posts within 24 hours.
Plagiarism & Academic Integrity:

Georgia Tech aims to cultivate a community based on trust, academic integrity, and honor. Students are expected to act according to the highest ethical standards.
Accommodations for Students with Disabilities:

If you require unique accommodation, contact the Office of Disability Services at 404-894-2563 or disabilityservices.gatech.edu as soon as possible to discuss your individual needs and obtain an accommodations letter.
Student-Faculty Expectations Agreement:

Mutual respect, acknowledgment, and responsibility are essential. Respect for knowledge, hard work, and cordial interactions will help build the desired environment.
Subject to Change Statement:

The syllabus and course schedule may be subject to change. Changes will be communicated via the Canvas announcement tool. Students must check discussion questions in the course, email messages, and course announcements to stay current in their online courses.