Introduction
Over the past years, the appearance of applications requiring or benefiting from (classical) artifical intelligence has accelerated. For example, electronic markets for the buying and selling of goods and services over the Web is a fast-growing, multi-billion-dollar segment of the world economy. Knowledge-based techniques for product recommendation, auctions, need identification, vendor selection, negotiation, agent communication, ontologies, business rules, and information integration are of rising interest and have started having practical impact on real Web e-markets.
This class covers the foundational theories (mostly) from the field of (classical) artificial intelligence that have made it possible to evolve to more “intelligent” applications. It will cover areas such as knowledge representation and reasoning (increasingly important through the semantic web effort of the w3c), problem solving, planning, and reasoning under uncertainty. For each of the subjects it will cover the underlying theories and provide a insight into practical applications using those techniques
Please note that though this class does not attempt to answer the grand old question of artificial intelligence: how to build an artificial intelligence. Its goal is to present methods found during this quest that have been surprisingly useful in practical applications.
A full syllabus draft is available as PDF.
People
Responsible Lecturer:
Responsible Assistants:
Literature
We will use: Russel, S., & Norvig, P. (2003). Artificial Intelligence: A Modern Approach. Upper Saddle River, NJ: Prentice-Hall.
Please note: The second edition of the book (published this year is vastly improved). There are some new chapters, that will be available in the library as we need them.
The Readings below are listed as the chapters in the second edition! A conversion table for the chapters can be found here.
Some of the new chapters in the second edition are available on-line!
Some additional papers and books. To be announced!
Prerequisites
Required: Finished Grundstudium
Of advantage:
- KV Datenbanksysteme
- KV Global verteilte und dynamische Anwendungssysteme
Time Table
To be revised!
Time |
Date |
Subject |
Readings (Chapters) |
Assignment |
Mo |
18.10. |
Introduction |
|
|
|
|
Part 1: Search |
|
|
|
|
Uninformed Search |
3 |
|
Mo |
25.10. |
Informed Search |
4 |
A1 out |
Mo |
1.11. |
Constraint Satisfaction |
5 |
|
Mo |
8.11. |
Search in Games |
6 |
|
|
|
Part 2: Knowledge Intensive Processing |
|
|
Mo |
15.11. |
Logic, First Order Logic |
skim 7, 8 |
A1 back |
Mo |
22.11. |
Reasoning in FOL (LP), Knowledge Representation |
9 & 10 |
A2 out |
|
|
Part 3: Learning |
|
|
Mo |
29.11. |
Tree inducers |
18 (18.1, 18.2, 18.3) |
|
Mo |
6.12. |
Bayesian Methods |
13 (20.1, 20.2) |
|
Mo |
13.12. |
No class |
|
|
Mo |
20.12. |
Spam filtering, fraud detection, house price estimation |
|
A2 back |
|
|
X-Mass break |
|
|
Mo |
10.01. |
Bayesian Belief Networks - Clippy and Co. |
14 (in particular beginning < 14.5) |
|
Mo |
17.1. |
Hidden Markov Models - Speech Recognition |
|
|
Mo |
24.1. |
Text Mining, Information Extraction |
|
A3 back |
Mo |
31.1. |
Machine Translation, Summary Session |
|
|
Mo |
7.2. |
Final exam |
|
|
Handouts
Slides
- Part 1: Search, etc.
- Introduction, Planning & Search
- Constraint Satisfaction Problems (CSP) (session 2)
- Games
- Part 2: Knowledge Intensive Processing
- Part 3: Learning
Assignments
- Assignment #1: Search, etc.
Sample Solutions, puzzle_reolon.jar, puzzle_krist.jar
- Assignment #2: Search in Games & Knowledge Representation sample solution: Part 1, Part 2a, Part 2b, Part 2c
- Assignment #3: Learning
Note for Part 1:
1) If the game reaches the same board status (A and B on same fields as before) but it is a different player's turn to move, this is definitively a different game status and the corresponding node in the decision tree has to be expanded.
2) This does not occur in the given assignment with n=6.
3) We think that point 1) may not occur for any n>2 and can be shown by induction (we omit the proof here).