Intelligent Systems 2II40 C2

1
Intelligent Systems (2II40)C2

• Alexandra I. Cristea

September 2004
2
Outline

• Intelligent agents
• Search

3
II. Intelligent agents

• Rational agent
• Agent its environment
• Example a simple agent
• Rationality?
• Task environment
• PEAS
• Properties of the task environment
• Agent properties

4
RecapTest Is the agent rational?

• A.- environment what is known about the
  environment?
• B.- task what is known about the given task?
• C.- machine resources what is known about
  the given machine resources?
• D.- percept sequence what is known about the
  precept sequence up to date?
• E.- agent actions
• F.- is there a performance measure?
• G.- After the questions above are answered, we
  have to check if the performance measure is
  maximized

5
Agent Rationality complete def.

• For each possible sequence task in a given
  environment, a rational agent should select an
  action that is expected to maximize its
  performance, given its resources, built-in
  knowledge and its performance measure.

6
II.5.A. Specifying the task environment PEAS
7
The agent and its environment
8
(No Transcript)
9
(No Transcript)
10
(Optional homework)
11
II.5.B. Properties of task environment
12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
II.6. Agent types
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
Conclusion agents

• Agent is something that perceives acts in
  an environment.
• Extra exercise find alternative definitions!
• A Rational Agent acts so that it maximizes the
  performance measure.
• A task environment includes performance measure,
  external env., actuators, sensors.
• Basic agent program design reflex, model/ goal/
  utility based, learning agents

27
Extra - Homework 2 part I

• Show that the reflex vacuum cleaner agent
  defined previously is rational (use the complete
  definition of rationality defined in C2).
• Develop a PEAS description for a robot soccer
  player and for a mathematicians theorem
  proving assistant give the proprieties of the
  task environment for each select a suitable
  agent design.

28
Outline

• Intelligent agents
• Search
• Uninformed
• Informed
• Heuristic
• Local
• Online
• Constraints satisfaction

29
Search applications

• Obvious
• Finding Olympic Games schedule on the Web.
• Finding the cheapest trip between here and Tokyo.
• A robot navigating an environment strewn with
  obstacles.
• A web-crawler indexing web pages
• Less Obvious
• Playing Chess
• Job Shop Scheduling
• Planning a party

30
Search ingredients

• Locations are called nodes
• Connections between nodes are called arcs
• Arcs can be directed so that they can only be
  traversed in one direction
• A collection of arcs and nodes is a net or a
  graph
• If every node in the net has a unique parent with
  one exception the net is a tree
• The unique node is the root and has no parents

31
A Directed Graph (DG)

• In fact a Directed Acyclic Graph (DAG)

32
Search

• If we want to search through this graph from S to
  F this graph can be viewed as a tree.

33
Search algorithms
34
General search

• Offline, simulated exploration of state-space
• Generating successors of already explored states
  (expanding)

35
Example traveling in Romania
36
General search example
Arad
37
General search example
Arad
Zerind
Sibiu
Timisoara
38
General search example
Arad
Zerind
Sibiu
Timisoara
Arad
Oradea
Fagarash
Ramnicu Valcea
Sibiu
Bucharest
39
Implementation of general search
40
States vs. nodes
Node
parent
State
Depth2
InFagarash
Fagarash
children
Sibiu
Bucharest
41
Strategy characteristics

• order node expansion ?
• parameters
• Completeness solution?
• Optimality best solution?
• Complexity
• Time max no. steps to solution
• Space nodes in memo
• parameters of complexity computation
• b max branching factor of search tree
• d depth of least-cost solution
• m max depth of state space

42
III.1.Uninformed search algorithms
43
Breadth-first search

• Expand shallowest node first
• Implementation FIFO queue

44
Breadth-first example
Arad
45
Breadth-first example
Arad
Zerind
Sibiu
Timisoara
46
Breadth-first example
Arad
Zerind
Sibiu
Timisoara
Arad
Oradea
Arad
Oradea
Fagarash
Ramnicu Valcea
Arad
Lugoj
47
Breadth-first example
Arad
Zerind
Sibiu
Timisoara
Arad
Oradea
Arad
Oradea
Fagarash
Ramnicu Valcea
Arad
Lugoj
Sibiu
Sibiu
Zerind
Timisoara
Zerind
Zerind
Zerind
Zerind
Bucharest
Sibiu
Sibiu
Sibiu
Sibiu
Craiova
Mehadia
Sibiu
Pitesti
Timisoara
Timisoara
Timisoara
48
Proprieties of breadth-first search

• Complete?
• Time?
• Space?
• Optimal?

49
Proprieties of breadth-first search

• Complete? Yes (if b, d finite)
• Time? O(bd1)
• Space? O(bd1)
• Optimal? Yes (if b, d finite cost/step1)
• Problem space!!

50
Depth-first search

• Expand deepest node first
• Implementation LIFO queue

51
Breadth-first vs Depth First

• Breadth-first

52
Depth-first example
Arad
53
Depth-first example
Arad
Zerind
Sibiu
Timisoara
54
Depth-first example
Arad
Zerind
Sibiu
Timisoara
Arad
Oradea
55
Depth-first example
Arad
Zerind
Sibiu
Timisoara
Arad
Oradea
Zerind
Sibiu
Timisoara

56
Iterative deepening search

• Depth first search with growing depth
• l allowed maximal depth in tree

57
Iterative deepening search example
l 0
Arad
58
Iterative deepening search example
l 1
Arad
59
Iterative deepening search example
l 1
Arad
Zerind
Sibiu
Timisoara
60
Iterative deepening search example
l 2
Arad
61
Iterative deepening search example
l 2
Arad
Zerind
Sibiu
Timisoara
62
Iterative deepening search example
l 2
Arad
Zerind
Sibiu
Timisoara
Arad
Oradea
63
Iterative deepening search example
l 2
Arad
Sibiu
Timisoara
Arad
Oradea
Fagarash
Ramnicu Valcea
64
Iterative deepening search example
l 2
Arad
Timisoara
Arad
Lugoj
65
Proprieties of iterative deepening search

• Complete? Yes (b,d finite)
• Time? (d1) db (d-1)b2 bd O(bd)
• Space? O((b-1)d)
• Optimal? Yes (b,d finite cost/step1)

66
Extra Homework 2 part II

• Compute the proprieties of the depth-first search
  (completeness, time -, space complexity,
  optimality). Hint some of the memory can be
  freed after usage.

67
Assignment 2

• MOT (moh) Testing Part 1
• ----------------------------------
• Perform this test as a group X, where X is your
  assigned group number
• create your own, personal teacher group account
  (group X) at http//e-learning.dsp.pub.ro/mot/
• Check what it all means at http//wwwis.win.tue.n
  l/acristea/HTML/USI/MOT/help/
• Create a new (domain) concept map of the domain
  model with concepts with all standard (domain)
  attributes
• Create a new (domain) concept map of the domain
  model with a different name with concepts with
  only three standard attributes
• Change the name of the root concept for each
  concept map in the domain model that you created
• Create three children for each root concept of
  the concept maps above
• Add more children to the concept maps, so that
  the depth becomes 3
• Change the default number of standard attributes
  to three (different from the ones chosen for the
  second CM) for the concepts of the first concept
  map, by editing it (edit concept map)
• Add two concepts to the first concept map, to
  check that the change works
• Select one concept and add extra attributes
  different from the standard attributes to it
• Add some keywords for at least two concepts
  (separate them with ) make sure that the
  concepts have at least one concept in common
• Calculate the relatedness relations and for at
  least one of the concepts, and check that the
  other concept with at least one keyword in common
  is found add it to the relatedness relations

68
Other info

• Check IS course 2II40 homepage for info on
• Evaluation of assignments
• Projects grouping deadlines
• http//wwwis.win.tue.nl/acristea/HTML/IS/