Advanced Topics in AI

Topic outline

• 

  General

  • In this course, you will be given an introduction to the basic ideas and techniques which are the basis of the design of intelligent computer systems. A specific emphasis lies on the statistical and decision-theoretic modeling paradigm. The techniques taught apply to a wide variety of artificial intelligence problems and serve as the foundation for further study in any application area.

    To make this course more fun and practical we will explore the different concepts using Pac-Man. With Pac-Man we will cover the topics of Search and Multi-Agent Search, Constraint Satisfaction Problems, Markov Decision Processes and Reinforcement Learning and build our own small intelligent agents. It's a hands-on way to understand how AI works in action and makes complex ideas easy to grasp.

    So, get ready to join Pac-Man and dive into the exciting world of AI!

  • Zoom Link

    Join the live sessions via:
    https://us02web.zoom.us/j/84812346196?pwd=b3RKbGFwMm9SY0RkQXRYTk9raVhtdz09
    
    Meeting-ID: 848 1234 6196
    Password: 586454

  • Discord

    To join the discord server, please use the following link: https://discord.gg/Y5XVSS5aPt

    You can find three text channels there:

    • general: for organizational matters and everything not related to contents
    • lectures-and-exercises: if you have questions regarding the topics discussed in the video lectures, the live sessions or the exercises and quizzes
    • programming-exercises: for all questions related to the programming exercises
  • Course Organization Page

    In this file, you will find all needed information on how to participate and successfully complete the course.

  • 	This course is offered in the xAIM's master's program. The course is co-financed by the European Commission instrument "Connecting Europe Facility (CEF) Telecom" through the project "eXplainable Artificial Intelligence in healthcare Management project" (2020-EU-IA-0098), grant agreement number INEA/CEF/ICT/A2020/ 2276680.

  • Copyright

    Course materials were adapted from the course CS188 at UC Bereley (http://ai.berkeley.edu).

    The materials can be used freely for educational purposes. Attribution to the original copyright holders must be retained as academically appropriate and a short notice should be given. For more details, check http://ai.berkeley.edu.

    Pac-Man is a registered trademark of Namco-Bandai Games, used here for educational purposes

• 

  Lecture 1: Introduction

  • Dear all,

    welcome to the course Advanced Topics in AI! Until the first live session, please look at the materials provided. Can you access the videos (see the file Introduction for the links). Can you download the lecture slides in the folder? Is everything working in the programming exercise?

    The topics of the recordings will also be discussed in the first live session and there we will also discuss organizatorical matters. If you have any questions beforehand, let us know via Discord (link see above).

    We are looking forward to an exciting semester with you!

  • Introduction Page

    Welcome to Lecture 1: Introduction! Embark with us on an exciting journey exploring the fascinating world of Artificial Intelligence (AI). In this lecture we will look at the representation of AI in pop culture, ask us what AI actually is, take a quick detour into the history of AI and come back to what AI can do today.

  • Lecture Slides Introduction Folder
  • Programming Exercise Folder

    The first programming exercise is meant to set you up for the upcoming ones. You don't need to submit it and cannot gain any bonus points. But it will help you get used to the structure.

• 

  Lecture 2: Search

  • Welcome to Lecture 2 - Search!

    In the file Search you can find initial information and the link to the video recordings. The folder Lecture Slides contains the slides. In the assignment Programming Excercise 1 you can find all the necessary material. Make sure to submit it on time! The folder Exercise contains only the exercise sheet at the moment, after the live session - where we will discuss it - you can also find the solution there.

  • Search Page

    Welcome to Lecture 2 - Search. This is the first step into the world of planning and problem-solving, where we learn about planning agents, how to translate a real-world problem into a formalized search problem and algorithms to solve these problems.

  • Programming Exercise 1 Assignment
  • Lecture Slides Search Folder
  • Exercise Search Folder
  • Small Step by Step Examples

  • Tree Search and Graph Search Quiz
  • DFS, BFS and UCS Quiz
  • A* Graph Search Quiz
  • Final Quiz for Lecture 2 - Search

    This is the Quiz for the contents of Lecture 2 - Search. You can retake the quiz however often you want before the deadline (October 25th, 18:00). Your last try will be evaluated. If you get at least 20 points out of 26, you will get a bonus point.

• 

  Lecture 3: Constraint Satisfaction Problems

  • Welcome to Lecture 3 – Constraint Satisfaction Problems.

    In the file Constraint Satisfaction Problems you can find initial information and the link to the recordings. The folder Lecture Slides contains the slides used in the recordings. The folder Exercise contains the exercise sheet and after the live session also the solution. There is no programming exericse in this unit.

  • Constraint Satisfaction Problems Page

    Lecture 3 focuses on Constraint Satisfaction Problems, a subset of Search Problems. We will learn how to solve them with general-purpose algorithms by using the structure of states.

  • Lecture Slides CSPs Folder
  • Exercise CSPs Folder
  • Small Step-by-Step Examples

  • Formalization Quiz
  • CSP Example Quiz
  • Improving Structure Quiz
  • Final Quiz for Lecture 3 - CSPs

    This is the Quiz for the contents of Lecture 3 - Constraint Satisfaction Problems. You can retake the quiz however often you want before the deadline (NEW: December 6th, 18:00). Your last try will be evaluated. If you get at least 15 points out of 20, you will get a bonus point.

• 

  Lecture 4: Multiagent Search

  • Welcome to Lecture 4 - Multiagent Search

    In the file Multiagent Search you can find initial information and the link to the recordings. The folder Lecture Slides contains the slides used in the recordings. The folder Exercise contains the exercise sheet and after the live session also the solution. In the assignment Programming Excercise 2 you can find all the necessary material.

  • Multiagent Search Page

    In Lecture 4 – Multiagent Search, we will learn how to account for other agents using Minimax and Expectimax. In the last parts of the lecture, we will look at how to describe agents’ preferences using utilities.

  • Programming Exercise 2 Assignment
  • Lecture Slides Multiagent Search Folder
  • Exercise Multiagent Search Folder
  • Small Step-by-Step Examples

  • Minimax and alpha/beta pruning Quiz
  • Expectimax Quiz
  • Utilities and Lotteries Quiz
  • Final Quiz for Lecture 4 - Multiagent Search

    This is the Quiz for the contents of Lecture 4 - Multiagent Search. You can retake the quiz however often you want before the deadline (November 22th, 18:00). Your last try will be evaluated. If you get at least 15 points out of 20, you will get a bonus point.

• 

  Lecture 5: Markov Decision Processes

  • Welcome to Lecture 5 – Markov Decision Processes

    In the file Markov Decision Processes you can find initial information and the link to the recordings. The folder Lecture Slides contains the slides used in the recordings. The folder Exercise contains the exercise sheet and after the live session also the solution. There is no programming exericse in this unit.

  • Markov Decision Processes Page

    In Lecture 5 – Markov Decision Processes we will learn how to maximize our utilities by determining the value of states and extracting the best action for each state. We will look at two methods to do that: value iteration and policy iteration, which consists of policy evaluation and policy extraction.

  • Lecture Slides MDPs Folder
  • Exercise MDPs Folder
  • Small Step-by-Step Examples

  • Formalization Quiz
  • Finite Horizons & Discounting Quiz
  • Value Iteration and Policy Iteration Quiz
  • Final Quiz for Lecture 5 - MDPs

    This is the Quiz for the contents of Lecture 5 - Markov Decision Processes. You can retake the quiz however often you want before the deadline (December 6th, 18:00). Your last try will be evaluated. If you get at least 15 points out of 20, you will get a bonus point.

• 

  Lecture 6: Reinforcement Learning

  • Welcome to Lecture 6 - Reinforcement Learning, our last lecture.

    In the file Reinforcement Learning you can find initial information and the link to the recordings. The folder Lecture Slides contains the slides used in the recordings. The folder Exercise contains the exercise sheet and after the live session also the solution. In the assignment Programming Excercise 3 you can find all the necessary material for that.

  • Reinforcement Learning Page

    In Lecture 6 - Reinforcement Learning we learn how to learn to act optimally in an MDP, where we don't know the Transition and Reward functions.

  • Lecture Slides RL Folder
  • Programming Exercise 3 Assignment
  • Exercise RL Folder
  • Small Step-by-Step Examples

  • TD Learning Quiz
  • Q-Learning Quiz
  • Approximate Q-Learning Quiz
  • Final Quiz for Lecture 6 - Reinforcement Learning

    This is the Quiz for the contents of Lecture 6 - Reinforcement Learning. You can retake the quiz however often you want before the deadline (January 10th, 18:00). Your last try will be evaluated. If you get at least 15 points out of 20, you will get a bonus point.

• 

  Final Exam

  • The final exam will be a quiz here on moodle (like the final quizzes for each topic). You can use the previous quizzes as preparation for the exam.

    The exam will take place on Thursday 29th of February. We will meet at 18:00 in Zoom (link is the same as always). There I will provide you with a pin code so you can access the exam. You will have 90min time to solve the exam.

    You should bring a calculator. This will be an open book exam, so you can bring all materials, which appear helpful to you. However, out of experience it might be more helpful to bring self-written short-summaries of the topics than all lecture materials.

  • Final Exam Advanced Topics in AI Quiz