Presentation is loading. Please wait.

Presentation is loading. Please wait.

1 Building Bayesian Networks COMPSCI 276, Fall 2009 Set 3: Rina Dechter (Reading: Darwiche chapter 5)

Similar presentations


Presentation on theme: "1 Building Bayesian Networks COMPSCI 276, Fall 2009 Set 3: Rina Dechter (Reading: Darwiche chapter 5)"— Presentation transcript:

1 1 Building Bayesian Networks COMPSCI 276, Fall 2009 Set 3: Rina Dechter (Reading: Darwiche chapter 5)

2

3 Queries: Different queries may be relevant for different scenarios

4 For other tools see class page

5 Other type of evidence: We may want to know the probability that the patient has either a positive X-ray or dyspnoea, X =yes or D=yes.

6

7

8 C= lung cancer

9

10 P(V=yes|E=yes) P(V=yes|E=no) =2 Define a cpt for V that satisfies this constraint

11

12

13

14

15

16 Is it correct?

17

18

19

20 Building email management network Step 1: define the variables: email characteristics Title, (values: any sequence of words.) sender-id, (values: # of id names) Recipients, #-of-recipients (Values, a sequence of id-names) topic, (values: a distribution over bag of words, or a set of key words) length, (values: natural numbers) time-sent : (time-of-week, time-of-day), (values: days of the week, time (discredized) time-read, (values: as above) current-time, (value: as above) max-reponse-time (value: as above) Evidence variables query

21

22

23 Variables? Arcs? Try it.

24 What about? A naive Bayes structure has the following edges C -> A1,..., C -> Am, where C is called the class variable and A1; : : : ;Am are called the attributes.

25

26

27

28 Learn the model from data

29 Learning the model

30 Try it: Variables and values? Structure? CPTs?

31

32

33

34

35

36

37 Skip if no time, this and the next 4 slides

38

39

40

41

42 Try it: Variables? Values? Structure?

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62 Variables? Values? Structure?

63

64

65

66

67

68 Try it: Variables, values, structure?

69

70

71

72

73

74

75

76

77

78

79 What queries should we use here?

80

81 WER (word error rate), BER (bit error rate) MAP (MPE) minimize WER, PM minimize BER… What do you think?

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

98

99

100 Building email management network Step 1: define the variables: email characteristics Title, (values: any sequence of words.) sender-id, (values: # of id names) Recipients, #-of-recipients (Values, a sequence of id-names) topic, (values: a distribution over bag of words, or a set of key words) length, (values: natural numbers) time-sent : (time-of-week, time-of-day), (values: days of the week, time (discredized) time-read, (values: as above) current-time, (value: as above) max-reponse-time (value: as above) Evidence variables query

101 Email network for a message Title Topic Sender-id recipients MaxWait Day-of-Week Time-of-Day Real-max-wait Time-now time-left #-of

102 Topic and title-subnetwork Topic Title w1w3w2w1 w100 tw2tw1 tw3tw4 Topics and titles will have a small number of categories Words are either present or not

103 Email network for a message Title Topic Sender-id recipients MaxWait Day-of-Week Time-of-Day Real-max-wait Time-now time-left #-of Topic

104 Slides 130-155 explain the domain. Read. Variables, values, structure?

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152 Applications 152 Linkage Analysis LOD Scores Computing Haplotypes

153 153 Two Loci Inheritance Recombinant 2 1 A B a b A a B b 34 a b A a b 56 A a B b

154 154 Bayesian Network for Recombination S 23m L 21f L 21m L 23m X 21 S 23f L 22f L 22m L 23f X 22 X 23 S 13m L 11f L 11m L 13m X 11 S 13f L 12f L 12m L 13f X 12 X 13 y3y3 y2y2 y1y1 Locus 1 Locus 2 P(e|Θ) ? Deterministic relationships Probabilistic relationships

155 155 L 11m L 11f X 11 L 12m L 12f X 12 L 13m L 13f X 13 L 14m L 14f X 14 L 15m L 15f X 15 L 16m L 16f X 16 S 13m S 15m S 16m S 15m L 21m L 21f X 21 L 22m L 22f X 22 L 23m L 23f X 23 L 24m L 24f X 24 L 25m L 25f X 25 L 26m L 26f X 26 S 23m S 25m S 26m S 25m L 31m L 31f X 31 L 32m L 32f X 32 L 33m L 33f X 33 L 34m L 34f X 34 L 35m L 35f X 35 L 36m L 36f X 36 S 33m S 35m S 36m S 35m Linkage analysis: 6 people, 3 markers

156

157

158

159

160

161

162

163

164

165

166

167

168

169

170

171

172

173

174

175


Download ppt "1 Building Bayesian Networks COMPSCI 276, Fall 2009 Set 3: Rina Dechter (Reading: Darwiche chapter 5)"

Similar presentations


Ads by Google