This is a list of corrections to An introduction to Bayesian Networks.

If you have further corrections, please mail fvj@cs.auc.dk


Semantic corrections

page x12 : in the mail address, 'iesd' should be substituted with 'cs'

page 14, the end of section 2.2.1, add: Definition. The Markov blanket of a variable A is the parents of A, the children of A and the variables sharing a child with A.

Note: If all variables in the Markov blanket for A are instantiated, then A is d-separated from the rest of the network (see exercise 2.3).

page 17, equation 2.6: It shall be 'P(ai | bj) = P(ai | bj, ck ) for all i,j,k'

page 197, add: To relate Bayesian networks to d-separation we require that any variable is independent of the rest given its Markov blanket.

page 292 : 'D1' rather than 'D2'

page 377: 'observations' rather than 'findings'. The concept 'finding' is defined on page 71

page 45 Table 3.8: The first column in the righthand table should be 0.24 0.76 rather than 0.4 0.6

page 531: 'Table 3.12' rather than 'Table 3.7' at both places

page 54: Remove the '-' in the formula for cross entropy (also called Kullbach-Leibler distance)

page 559: 'Table 3.12' rather than 'Table 3.7'

page 62: remove the '-' in the formula for cross entropy

page 62: square the difference in the definition of euclidean distance

page 628: 'Ai' rather than 'Aj'

page 71, Table 4.3: The column with three b1's shall be b1, b2, b3

page 713 : 'A not in state' rather than 'A in state'

page 726 : In the righthand equation 'P(U, e )' and 'P(U | e )' shall be swapped

pages 75-93, Figures 4.3, 4.6, 4.7, 4.8, 4.26: The links should be undirected

page 76, the first equation: the table t** should have index V

page 76, the second line of equations: V and W should be interchanged; the first table should have **

page 94, Table 4.7: The table is multiplied by 100 rather than ten

page 96, Exercise 4.11 ii): The formula should be P(D | a ) = +B,CP(D | B, C ) P(B | a ) P(C | a )

page 1023-9: The variable is 'J' rather than 'E'

page 10413: 'Table 5.1' rather than 'Table 4.4'

page 110 the local conflict formula: divide by P(e1, ..., en)

page 112-113: The variable is 'G' rather than 'J'

page 113 Table 5.4: In the fraction of the righthand column 'e' ' rather than 'e'

page 12317: "while scanning has ..." shall rather be "while a combination of all tests has ..."

page 1378 : 'U(drill, O )' rather than 'U(drill | O)'

page 1527: 'U(a, ...)' rather than 'U(A, ...)'

page 1526: a paranthesis is missing. alpha is also multiplied on beta.

page 15310: Opt( ....) shall be conditioned by e


Typos, misspellings etc.

page x6: Larraņaga rather than Larra^naga

page 315: 'The reason being is ...' remove 'is'

page 5, caption to Figure 1.5: 'workstation' rather than 'work station'

page 1514: 'with' rather than 'wilth'

page 192: 'probabilities' rather than 'probabilitiess'

page 2313: 'sets' rather than 'clusters'

page 292: 'said' rather than 'sid'

page 569:'P(Word)' rather than P(Word'

page 7117 : The vector shall read (0, ..., xi, 0, ..., 0, xj, 0, ..., 0)

page 7614 : 'from V to W' rather than 'from V and W'

page 81: in figure 4.11 the separator between BCF and CDG is C rather than D

page 96 Exercise 4.10: Subquestion iii) is missing

page 1506 : 'U(t, h)' rather than 'U(h, t)'