About Us
Organization
Exercises and Exam
Content
Lectures
Artificial Intelligence 0. About this Course Who Will Do What, How, When, and What For
J¨org Hoffmann
Wolfgang Wahlster
Summer Term 2016
Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
1/33
About Us
Organization
Exercises and Exam
Content
Lectures
Agenda
1
About Us
2
About the Organization
3
About the Exercises and Exam
4
About the Content
5
About My (Hoffmann’s) Lecturing . . .
Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
2/33
About Us
Organization
Exercises and Exam
Content
Lectures
Foundations of Artificial Intelligence (FAI) Group
What? Basic research in AI. Since when? April 2012. Who? Prof. J¨ org Hoffmann (lead):
[email protected] Dr. Alvaro Torralba (postdoc):
[email protected] Daniel Gnad (PhD student):
[email protected] Marcel Steinmetz (PhD student):
[email protected] B¨arbel Zitzmann (secretary):
[email protected]
Where? E1 1, 3rd floor. http://fai.cs.uni-saarland.de/
Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
4/33
About Us
Organization
Exercises and Exam
Content
Lectures
DFKI and FAI: Research DFKI: (BIG) Human-Computer Interaction. Computer Graphics, Virtual Reality. Semantic Web. Language Processing. ... Applied and basic research. Big integrating projects with industry involvement. FAI: (SMALL) Automatic Planning, General Game Playing, related areas. Combinatorial search, heuristic function design and analysis, declarative problem description languages (logics), . . . Basic research. Algorithm design and analysis. Small projects. Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
5/33
About Us
Organization
Exercises and Exam
Content
Lectures
Who Does What in This Course: FAI Group & DFKI
J¨org Hoffmann and Wolfgang Wahlster: Lectures. Alvaro Torralba, Daniel Gnad, Marcel Steinmetz: Overall course organization, tutorials supervision Hoffmann’s part. J¨org Baus, Boris Brandherm: Tutorials supervision Wahlster’s part. B¨arbel Zitzmann: Contact for administrational issues like HISPOS and certificates (“Schein”) at the end in case of need.
Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
6/33
About Us
Organization
Exercises and Exam
Content
Lectures
Who Does What in This Course: Our Tutors
Cosmina Croitoru:
[email protected] Daniel Heller:
[email protected] Akram El-Korashy:
[email protected] Bj¨orn Mathis:
[email protected] Lukas Schaal:
[email protected] Julia Wichlacz:
[email protected]
Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
7/33
About Us
Organization
Exercises and Exam
Content
Lectures
Who Does What in This Course: You
. . . and who are you?
More precisely: Which course of studies (Studiengang) do you follow? Informatics MSc? Informatics BSc? Bioinformatics? CuK? Cybersecurity? Information systems? Media Informatics? Visual Computing? Computational Linguistics? Mathematics? Other? (Which?)
Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
8/33
About Us
Organization
Exercises and Exam
Content
Lectures
Where, What, When: Lectures Where? G¨ unter-Hotz-H¨orsaal (E2 2). What? (rough classification!!) Basics/Principles: J¨org Hoffmann. Applications: Wolfgang Wahlster. When? Mondays 10:15–11:45, Tuesdays 16:15–17:45. Some exceptions. April/May/June: Hoffmann. July: Wahlster. See the overview on the course web page.
Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
10/33
About Us
Organization
Exercises and Exam
Content
Lectures
Web Resources Course Web Page: http://fai.cs.uni-saarland.de/teaching/summer-16/ai.html
Basic facts about the course. Lecture slides hand-outs. Not updated at all, apart from the lecture slides! Course Moodle Pages: https://fai-lecture.cs.uni-saarland.de/login/index.php
Everything apart from the lecture slides! Registration (see also next section). Announcements, tutorial groups, exercise sheets, discussion forum for technical questions, . . . Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
11/33
About Us
Organization
Exercises and Exam
Content
Lectures
Your Questions
Technical questions about course content/exercises: “Technical Discussion” in Moodle pages. (Read by everybody: All students, the whole AI’16 team.) Also, of course: Your tutor, if it’s about the exercises. Other questions: J¨org Hoffmann. Come to the front directly after the lecture.
Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
12/33
About Us
Organization
Exercises and Exam
Content
Lectures
You Must Register! You must register for the tutorials! Go to Moodle pages and “Create new account”. → Attention: You must use a valid email address ending with “.uni-saarland.de”. (If you have no such email address, contact Alvaro Torralba.) Go to “Artificial Intelligence (Summer 2016)”; when prompted, enter key “UDS-AI-16”. Go to “Tutorial groups” and choose an available tutorial group. → Size limit is 30, first come first served. Once registration is closed, you can switch between tutorial groups only if you find an exchange partner in the respective other group. In such case, contact the tutors of both groups involved. Registering for the exam (6= tutorials): HISPOS. Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
14/33
About Us
Organization
Exercises and Exam
Content
Lectures
You Must Register! Ctd. Tutorial groups: The time slot, place, and tutor for each group will be listed in the Moodle pages. Registration Timing: Registration will open tomorrow, April 19, at 12:00. We expect to close registration on Thursday, April 21, midnight. But that may be subject to changes. Pay attention to the announcements in the Moodle pages. Student groups: You may solve the exercises in groups of up to 3 students. All authors must be from the same tutorial group. The same group must address both, the paper and the practical exercises. Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
15/33
About Us
Organization
Exercises and Exam
Content
Lectures
Exercises TWO different kinds of exercises: Paper exercises: Understand and apply concepts from the course. Practical exercises: Get some experience with AI modeling languages and tools. Paper Exercises: Apply concepts and algorithms to examples, lead simple proofs. 1-week intervals for hand-out/submission. Hand-out in Moodle, submission deadline stated on each sheet. Practical exercises: Model given problems in AI problem-solving formalisms, solve with off-the-shelf tools. Models checked manually by the tutors, graded based on correctness. 1-week or 2-week intervals depending on topic and lecture timing. Hand-out in Moodle, submission deadline stated on each sheet. Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
16/33
About Us
Organization
Exercises and Exam
Content
Lectures
Exercises: Organization Hand-out: Moodle, Tuesday week X Paper exercises: 10 exercise sheets, each 10 points. First sheet 3.5. Practical exercises: First sheet 17.5. Submission: Before Tuesday lecture, week Y Paper exercises: Week Y=X+1. Paper solutions only! Collected at the front of lecture hall. Staple solutions, write names at top. (Otherwise, 3 points subtracted from sheet.) Practical exercises: Week Y=X+1 or Y=X+2 (deadline stated on sheet). Submission in Moodle. Tutorial groups: Week Y+1 Participation not obligatory. But highly recommended! → If you have successfully presented N ≥ 1 exercises, you get 3 extra points. (But no additional extra points for N ≥ 2.) Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
17/33
About Us
Organization
Exercises and Exam
Content
Lectures
Exams The exams will consist of exercises similar to the paper exercises. Exams rules: Exam qualification: ≥ 50 points in paper exercises, AND ≥ 50 points in practical exercises. ATTENTION! If you want to attempt the course again in later years, you must qualify anew.
Open book. Any paper material allowed. No phones or computers. Exam vs. re-Exam: Each is a separate attempt to pass this course. Both exams taken =⇒ better score counts. ATTENTION! Once you pass the course, you can NOT improve your grade in later years. The re-exam (if you already passed the exam) is your only chance to do so.
Exam preparation: (Instead of regular lecture on Monday, July 18) Lists exam-relevant parts of course; example exercises. Opportunity to ask questions. Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
18/33
About Us
Organization
Exercises and Exam
Content
Lectures
Exams Dates & Locations Exams Dates: Exam: Monday, July 25, 10:00-12:30, G¨ unter-Hotz-H¨orsaal AND HS002 (students distributed alphabetically). Re-Exam: Monday, October 10, 13:30-16:00, G¨ unter-Hotz-H¨orsaal AND HS002 (students distributed alphabetically). Exams INSPECTION Dates: Exam Inspection: Tuesday, August 2, 14:00-16:00, E1 1 room 3.06. Re-Exam Inspection: Friday, October 14, 14:00-16:00, E1 1 room 3.06. → This is a big course, and individual inspection dates are infeasible. You know the inspection date & time several months in advance. If you want to inspect your exam then you should mark your calendar. Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
19/33
About Us
Organization
Exercises and Exam
Content
Lectures
Course Content Artificial Intelligence (AI) → What kind of content do you expect? “AI” dimensions: (More in Chapter 1 tomorrow) (i) Thinking vs. acting. (ii) Human-like vs. rational. The modern AI pick: (i) Acting: More practice-oriented, and not all activity requires “thinking”. (We’ll do some “thinking” anyhow, see slide 23.) (ii) Rational: Try to act optimally, using whatever computational methods are suitable for that purpose. (Not necessarily imitating human decision making.) → We will talk about methods for making intelligent action choices. Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
21/33
About Us
Organization
Exercises and Exam
Content
Lectures
And here’s what you’ve all been waiting for . . .
AlphaGo = search + neural networks → We do search here (the basics). Neural networks/deep learning basics are covered in the Machine Learning courses. Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
22/33
About Us
Organization
Exercises and Exam
Content
Lectures
Course Outline Blue: Hoffmann; Red: Wahlster. Keyword? Introduction General Problem Solving Intelligent Agents Classical Search Adversarial Search Constraint Satisfaction Problems Propositional Reasoning First-Order Reasoning Planning Knowledge Representation Applications
Hoffmann and Wahlster
Content? Clarify the background Illustration of a key idea in AI Establish some basic concepts How to find routes, solve puzzles, find bugs in software, . . . ? How to solve games? How to schedule sports events, car manufacturing, . . . ? How to “think” rationally? A more powerful framework for “thinking” How to describe and solve all action-choice problems? Practical & extended forms of reasoning Applications at DFKI
Artificial Intelligence
Topic? Introduction Introduction Introduction Search Search Search Logics Logics Search & Logics Logics Applications
Chapter 0: About this Course
23/33
About Us
Organization
Exercises and Exam
Content
Lectures
Course Material Most of (Hoffmann’s part of) the course is based on Artificial Intelligence: A Modern Approach, Third Edition Stuart Russell and Peter Norvig (RN) . . . but this book is not our “Bible”: It’s great to get intuitions on the basics of what we’ll discuss. The book basically is “broad but shallow”: An immense breadth of AI sub-areas is being covered. Thus the book does not cover many important recent developments, and it often lacks formality. → We focus on a smaller range of areas, treated in more depth. → RN is the basis of many AI courses out there, e.g., Norvig and Thrun’s famous Udacity course (https://www.udacity.com/course/cs271). These courses partially overlap with the present one. They can provide useful additional background/explanations, but differ in many details. → The “ground truth” for this course are the post-handouts. Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
24/33
About Us
Organization
Exercises and Exam
Content
Lectures
Chapter Structure, Hoffmann’s Part Chapters, each with several sections. (E.g.: Chapter 0, About the Lectures section) Each chapter consists of an Introduction section, several technical sections, and a Conclusion section. Features of Introduction: Applications: What kind of problem can we address with this? Intuitions: What is the basic idea? Overview: What will we do in this chapter, and why.
Features of Conclusion: Summary: Main keywords and punchlines of the chapter. Topics not covered: Briefly points out that we really do only the very basics of the universe behind each chapter. Reading: Mainly points out which chapters of RN are relevant, and how they differ from the content as presented. Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
25/33
About Us
Organization
Exercises and Exam
Content
Lectures
Course Prerequisites and Aims Prerequisites: Algorithms: Solid knowledge. Complexity theory: Basic knowledge (NP-hardness etc.) Aims: At the end of the course, you should . . . . . . be familiar with prominent research areas in search and logics. . . . understand, and be able to apply, the basic concepts and algorithms of these (main ability required for successful exam!). . . . have the basis to specialize in, and work on, an AI research subject at FAI or DFKI. → BSc, HiWi jobs, MSc, PhD
Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
26/33
About Us
Organization
Exercises and Exam
Content
Lectures
Questionnaire Question! How many scientific articles (6-page double-column “papers”) were submitted to the 2016 International Joint Conference on Artificial Intelligence (IJCAI’16) in New York City? (A): 7 (B): 811 (C): 1996 (D): 2296 → Answer (D) is correct. (Previous year, IJCAI’15, answer (C) was correct . . . )
Questionnaires: At end of section/at start of 5 min break. You get 2 minutes/5 minutes. You’re free to make noise (e.g., discuss with your neighbors). Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
28/33
About Us
Organization
Exercises and Exam
Content
Lectures
More Generally: My Questions to You When will I ask them? In questionnaires. At various points during the lectures. We’ll do many examples together. Why do I ask them? They give you the option to follow the lectures actively. They allow me to check whether or not you are able to follow. How will I look for answers? “Streber syndrom”: 3 students answer all the questions, N − 3 sleep. If this happens, I may resort to picking students randomly. → There is nothing to be ashamed of when giving a wrong answer! You wouldn’t believe the number of times I got something wrong myself (I do hope all bugs are removed now, but . . . ) Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
29/33
About Us
Organization
Exercises and Exam
Content
Lectures
Slides Availability “Gotcha! I’ll just look up the answers on the hand-outs!” → It’s not gonna be that easy: Pre-Handouts:
Post-Handouts:
Without answers to questions.
With answers to questions and details for examples.
Without details for examples.
Corrections, where applicable.
Availability:
Availability:
1 day before chapter begins.
Day the chapter ended.
1 slide/page.
1 slide/page.
4 slides/page (“-4up”).
4 slides/page (“-4up”).
Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
30/33
About Us
Organization
Exercises and Exam
Content
Lectures
Post-Handouts → The post-handouts contain full detail, including all examples. You will be able to read them basically like a book. To be absolutely clear: Rest assured that all that’s relevant will be in the post-handouts. (= All the technical details, as contained in my presentation slides but not the pre-handouts.) Except of course illustrations to answer questions from the audience. Taking a few notes might be useful still . . . → I’m not sure how useful it is to print the pre-handouts. I’m giving you the option to, but would recommend taking notes on plain paper. (The slide numbering is consistent across all versions of the slides, for easy reference.) Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
31/33
About Us
Organization
Exercises and Exam
Content
Lectures
What You Should Do I would like to advocate exactly 2 options: (A) Read a “book” (post-handouts as well as RN and/or web sources), not coming to the lectures. (B) Come to the lectures, pay attention to my questions, interrupt me whenever you have a question. What I would like to avoid is the 3rd option: (C) Come to the lecture and sleep. Why am I saying this? Both (A) and (B) are valid from my point of view. The only advantage of (B) over (A) is that books don’t interact with you (they don’t answer your questions). → If we don’t interact, what’s the point of being here? Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
32/33
About Us
Organization
Exercises and Exam
Content
Lectures
And Now . . .
That’s It! Enjoy the course!
Hoffmann and Wahlster
Artificial Intelligence
Chapter 0: About this Course
33/33