EXPERT SYSTEMS FOR SPECIES DISTRIBUTION MODELING DEFINING EXPERT

EXPERT SYSTEMS FOR SPECIES DISTRIBUTION MODELING

DEFINING EXPERT SYSTEMS § A computer program that uses knowledge and reasoning techniques to solve problems that normally require the abilities of human experts. § Software that applies human-like reasoning involving rules and heuristics to solve a problem. § An expert system is software that attempts to provide an answer to a problem, or clarify uncertainties where normally one or more human experts would need to be consulted Nieuwenhuis & Toxopeus (2010) 2

EXPERT KNOWLEDGE § At the center sit the human experts. Bird watchers at work Salamandra Nieuwenhuis & Toxopeus (2010) 3

RULE BASED § Expert systems are rule based: If height > 1700 then presence = unlikely If slope > 40 degrees then presence is unlikely If vegetation = mixed forest then presence = low Nieuwenhuis & Toxopeus (2010) 4

BAYESIAN STATISTICS § Bayesian theorem: § E = evidence § H = hypothesis § P(H) = prior probability § P(E) = marginal probability: the probability on E § P(E | H) = conditional probability: the probability on E given H § P(H | E) = posterior probability: the probability on H given E Nieuwenhuis & Toxopeus (2010) 5

WHICH BOWL IS THE COOKIE FROM? § Bowl #1 has 10 chocolate and 30 plain cookies, bowl #2 has 20 of each. Fred picks a bowl at random, and then picks a cookie at random. The cookie turns out to be a plain one. How probable is it that Fred picked it out of bowl #1? § Answer: H 1 == bowl #1, H 2 == bowl #2. The bowls are identical, thus P(H 1) = P(H 2), the two add up to 1, so both are equal to 0. 5. The event E is the observation of a plain cookie. From the contents of the bowls, we know that P(E | H 1) = 30 / 40 = 0. 75 and P(E | H 2) = 20 / 40 = 0. 5. § Then: Nieuwenhuis & Toxopeus (2010) 6

BAYESIAN EXPERT SYSTEM § The inference engine combines the expert knowledge of the rules with the bayesian statistics § This gives a measure of confidence in the result (classification) Nieuwenhuis & Toxopeus (2010) 7

ACQUIRING EXPERT KNOWLEDGE § Fieldwork (building up knowledge yourself) § Literature § Experts in the field (interviews) Nieuwenhuis & Toxopeus (2010) 8

IN PRACTICE: CONDITIONAL PROBABILITIES AKA EXPERT KNOWLEDGE Nieuwenhuis & Toxopeus (2010) 9

IN PRACTICE: OUTPUT PREDICTION AND CONFIDENCE Nieuwenhuis & Toxopeus (2010) 10

IN PRACTICE: CHECK Nieuwenhuis & Toxopeus (2010) 11