Computer Laboratory > Teaching > Course material 2008–09 > Artificial Intelligence I


Artificial Intelligence I

Principal lecturer: Dr Mateja Jamnik
Taken by: Part IB

  • Syllabus

  • Lecture notes ([for printing 4up)] [slides]):
    • Introduction to AI and agents: [4up pdf] [slides pdf]
    • Problem solving by search, games (adversarial search) and constraint satisfaction problems: [4up pdf] [slides pdf]
    • Problem solving by planning; knowledge representation and reasoning: [4up pdf] [slides pdf]
    • Introduction to machine learning with artificial neural networks: [4up pdf] [slides pdf]

  • Other interesting and helpful links:
    • Modern rendition of (anti) Turing Test in Captcha with images: article
    • Intelligent agents - cars that think: video clip (part 1)
    • More about A* search algorithm including tracing in the graph from wiki.
    • This will demonstrate the use of A* search algorithm for path finding. You can set your own obstacles on the grid: applet
    • This applet will demonstrate the IDA* search algorithm for the sliding-tile puzzles: applet
    • Try out and understand the MINIMAX and ALPHA-BETA PRUNNING algorithms by tracing the steps in the search tree: applet
    • A tutorial about Constraint Satisfaction Problems (CSPs) with n-queen problem for n=4: web page tutorial
    • A sudoku solver using CSPs to solve the puzzle; it also shows how constraints and heuristics eliminate values from the domains for each square (i.e., variable). There is also an explanation of the heuristics used: applet of the solver, explanations
    • AI planning used in consol-based games: paper by J. Orkin
    • Semantic nets are used in Wordnet, a lexical database for English language, and MultiNet, a multilayer knowledge representation semantic network: Wordnet, MultiNet
    • Backpropagation worked example from the lecture (part 1 has the setup for the problem, initial values, a summary of a backpropagation algorithm, forward pass, and errors for the output node, part 2 has errors for hidden nodes, update of weights for all links, and one more cycle-epoch): part1, part2 (need Raven account)
    • Worked examples from last lecture (search techniques, A* search, planning Sussman Anomaly): slides (need Raven account)

  • For supervisors: please use past exam questions (back until 2002/2003 are the most relevant) for practice. You can also pick some of the exercises from the end of each relevant chapter of the textbook. Plus, the questions about topics that students find tricky are a good starting point.

  • Past exam questions

  • Feedback: Please provide feedback through the online lecture course feedback form.