January 12 (Tue), 2021 Introduction to the Course: syllabus, objectives, course structure, and topics to be covered. Bayesian networks and decision graphs as a topic area with artificial intelligence. Students introduce themselves. Items 1-6 and 8 in the Class agenda were covered.
January 14 (Tue), 2021 Example of plausible reasoning and causal graphs: icy roads, wet grass, earthquake. Reasoning in the evidential and causal direction. Intercausal reasoning and explaining away. A first look at the Hugin Bayesian network and influence diagram tool. Reasoning in the evidential and causal direction. Detailed small example of computation of certainties in a compositional system. Problems with compositional system. All items in the Class agenda were covered.
January 19 (Tue), 2021 HW1: Do exercises 2.1 through 2.11 [J], due on Thursday, 2020-01-28. The Chernobyl example and the importance of structure in a causal model. Causal networks and d-separation through most of section 2.2 [J]. Items 1-7, most of 8, 9, and 11 in the Class agenda were covered.
January 21 (Thu), 2021 D-separation with examples. Lauritzen's d-separation algorithm using moralized ancestral graphs. (A proof of correctness has been posted on the course website under "Lecture Notes.") Interpretations of probability: classical, frequentist, and Bayesian (subjective), started. Items 1-9 and 11 in the Class agenda were covered; item 10 was started.
January 26 (Tue), 2021 Survey: syllabus awareness and HW1 progress. HW2 assigned, due on 2021-02-04 (Thursday): Exercises 1.7, 1.11, 1.12, 1.13 [J]. Probability prerequisites; potentials; an algebra of potentials (Ch.1 [J]). Items 1-8 and 10 in the Class agenda were covered.
January 28 (Thu), 2021 Causal and Bayesian networks. Neapolitan's [1990] definition of Bayesian networks and its equivalence to Jensen and Nielsen's one. The local Markov condition. The chain rule for Bayesian networks. All items in the Class agenda were covered.
February 2 (Tue), 2021 HW3 assigned: exercises 2.12, 2.13, 2.14, 2.17, 2.18, 2.19 [J], due on 2021-02-11 (Thursday). Please check the errata page, because there is a correction to the statement of exercise 2.19: variables A and B need to be exchanged. Discussion of HW1. All items in the Class agenda were covered.
February 4 (Thu), 2021 Variable elimination. Munin. Probabilistic graphical models and their advantages. Ch.2 [J] slides from the authors completed. Three computational problem: belief assessment (update), MPE, MAP. All items in the Class agenda were covered.
February 9 (Tue), 2021 Bucket elimination for belief assessment (update) with detailed example. Discusion of variable elimination and bucket elimination. All items in the Class agenda were covered; item 8 partly.
February 11 (Thu), 2021 HW4 assigned: Exercises 2.21, 2.22, 2.23 [J] due 2021-02-18; These exercises require Hugin; See agenda for details. Non-serial dynamic programming. All items in the Class agenda were covered.
February 16 (Tue), 2021 Review of HW3. Building models (Ch.3 [J]) started. All items in the Class agenda were covered.
February 18 (Thu), 2021 Building models (Ch.3 [J]) continued, to the end of "Catching the Structure." All items in the Class agenda were covered, except that the stratum method was not covered.
February 23 (Tue), 2021 HW5 assigned, due on Thursday, 2021-03-04. The stratum method. Determining the conditional probabilities started. All items in the Class agenda were covered,
March 2 (Tue), 2021 Review of HW4 (exercise 2.23 especially). The coins and bell example. All items in the Class agenda were covered,
March 4 (Thu), 2021 Determining the conditional probabilities through the poker example. All items in the Class agenda were covered,
March 9 (Tue), 2021 HW6 assigned: exercises 3.10, 3.12, 3.13 [J], due on 2021-03-18 (Thursday). All items in the Class agenda were covered,
March 11 (Thu), 2021 Midterm Exam. Here is the Class agenda.
March 16 (Tue), 2021 Midterm exam post-mortem. Discussion of issues concerning the exercises in HW6. All items in the Class agenda were covered.
March 18 (Thu), 2021 HW6 due date changed to 2021-03-23 (Tuesday). HW6 changed: Exercises 3.10, 3.12, 3.13 (part i-iii only) [J]. Two detailed examples of successful Bayesian networks: BOBLO (BOvine BLOod Typing) and CHILD (a system to support the diagnosis of "blue babies"). Building models: undirected relations (constraints). All items in the Class agenda were covered.
March 23 (Tue), 2021 Undirected relations (logical constraints). Expert disagreement and adaptation basics. Sensitivity to evidence and to parameters (basics). Continuous variables with the extended angina example. Temporatl clones and DBNs (started). All items in the Class agenda were covered.
March 25 (Tue), 2021 Discussion of graduate student assignment. Some suggestions: a presentation based on the CHILD network; a report and presentation based on the BOBLO network, or a presenation based on a paper from the special section of issue 17, number 1 (2021) of _Integrated Environmental Assessment and Management_, which is available online through the university library. Dynamic Bayesian networks (sectin 3.3.7 [J07]). Propagation in Bayesian networks and the Junction tree method, following the presentation in ch.4 [J96], available on Blackboard under "Course Content," started. The items in the Class agenda were covered.
April 1 (Thu), 2021 Review of HW6. Argument for the choice of parameter values that guarantee stability for the model of exercise 3.10. Propagation in Bayesian networks and the Junction tree method, following the presentation in ch.4 [J96], available on Blackboard under "Course Content," up to the beginning of section 4.4. The items in the Class agenda were covered.
April 6 (Tue), 2021 HW7 assigned: exercises 3.16, 3.21 (second part only; estimate the probabilities using the Noisy-Or assumption only), 3.7, 3.29 [J], due on 2021-04-13. Discussion of required extra work for graduate credit. Graduate students and undergraduates taking the course for graduate credit need to email the instructor by Thursday, April 8, with a proposal. Belief updating: the junction tree method, through section 4.4 [J96]. The items in the Class agenda were covered.
April 8 (Thu), 2021 The junction tree method (sections 4.1-4.5 [J96]) completed. The items in the Class agenda were covered.
April 13 (Tue), 2021 The extra work for graduate students is due on April 26, 2021. Stochastic simulation: forward sampling, Gibbs Sampling, likelihood weighting (Section 4.5 [J96], Section 4.8 [J]), loopy belief propagation (Section 4.9 [J]), recursive conditioning (Section 4.7 [J]). The items in the Class agenda were covered.
April 15 (Thu), 2021 Graphical Languages for Specification of Decision Problems (ch.9 [J]) through Section 9.2 [J]. The items in the Class agenda were covered.
April 20 (Tue), 2021 The final exam will be a take-home exam. Graphical Languages for Specification of Decision Problems (ch.9 [J]) through Section 9.3 [J] (Decision Trees). The items in the Class agenda were covered.
April 22 (Thu), 2021 Final exam read and briefly discussed. The exam is in a departmental dropbox entry. It is due on 2021-05-04 at 1500 via dropbox. The extra work for graduate students is due on 2021-04-26, also via dropbox. Graphical Languages for Specification of Decision Problems, completed (ch.9 [J]). The items in the Class agenda were covered. End of course.