Michael P. Verdicchio

I recently included a section on d-Separation in my most recent talk on causality, but I wanted to give it its own post. Before defining it formally, a brief history is given here from Richard Scheines’s page at CMU (http://www.andrew.cmu.edu/user/scheines/tutor/d-sep.html).

Read the rest of this entry »

This semester we’ve covered a number of topics in Sungwook Yoon’s Statistical Relational Learning reading group. My turn came a couple of weeks ago and I presented Bayesian Logic Programming. It is essentially a methodology which combines the structural conveniences of Bayesian networks and the theorem proving aspects of logic programming. The chapter in the Getoor/Taskar text was not especially liked in the group as it lacked, among a number of things, specific examples of its use and advantages; in other words, we didn’t know WHY we should use such a framework, only that it was an interesting hybrid of two seemingly separate methodologies. Nonetheless, my slides are below.

Slides (ppt)

In the last Computational Systems Biology lab group seminar I presented the topic of causality. It was essentially a survey of the first two chapters of Judea Pearl’s book, Causality. The slides for the talk can be found at the link below.

Causality Seminar Slides (ppt)

Here is the most recent talk in my Bayes nets study. With it I will have wrapped up what I want to cover from Heckerman’s tutorial.  It concerns causality and then a gentle introduction to dynamic Bayes nets.
(ppt)

In my Statistical Relational Learning reading group I have a few slides to chirp in with regarding joint probabilities in Bayesian Networks.  The first slide show has a brief synopsis of what a Bayes net is, and then has a simple probability factorization example.  The second set of slides just lists some different application areas where Bayes nets are in use.  It focuses on biologically related areas, and then summarizes one paper.  Another student is giving the bulk of the talk.

Bayesian Network Example (ppt)

Bayesian Networks in Use (ppt)

Busy times and much has transpired since the last Bayesian networks talk. We’ve now completed parameter learning and had our first talk on structure learning. Here are the materials:

Talk 5 (pdf) – Long talk on parameter learning–we got through page 8.

Talk 6 (pdf) – Short talk finishing the previous talk and completing parameter learning.

Talk 6 Handout (pdf) – This handout summarizes many of the definitions, theorems and lemmas from parameter learning. The reference numbers and notation correspond to Neapolitan’s text.

Talk 7 (ppt) – This presentation gave an overview of structure learning via local search. It is lacking in examples and explicit theory, but examples will be shown more as we come to causality, and the theory is based on the theory presented in parameter learning.

Hopefully after the semester I’ll get one big Latex version of all of these.

So my seminar last Monday (not today) went horribly.  At first I had the simple goals of presenting on 1) learning parameters of Bayesian networks and 2) learning structure of Bayesian networks.  In my quest I quickly discovered that these topics are as deep as you’d like them to be, and that it takes 2/3 of the Neapolitan text book to cover it all.  Instead of doing a nice, concise presentation that covers only the essential details, I dragged my audience through my own tumultuous learning process.  Even though I’ve posted notes from bad talks before, these ones aren’t going up.

Next week I will give it another shot.  While my background knowledge must be extensive and airtight, my talk should be just an overview of important concepts.  I look forward very much to posting notes from the talk that don’t suck so much.

Over the summer and this semester I have been endeavoring to teach myself about Bayesian networks and their usability in modeling biological systems, and specifically gene interactions. The task has proved to be a difficult one, as my approach has not been very structured. It started with going through the 1995 tutorial by David Heckerman, but with each step I’ve had to make side journeys to fill in gaps in background knowledge.

So far in our lab seminar series (http://sysbio.fulton.asu.edu) I’ve covered the background in probability theory and in graph theory (Markov chains/blankets/equivalence, faithfulness, d-separation, etc.), and have begun discussing how to actually learn Bayesian networks from data. The last thing discussed was an intro on leaning posterior probability distributions, which will lead into learning network topology, and finally learning both.

I am posting my slides/notes from the four talks I’ve given, but take them only as an illustration of my progress and not as a good resource. For good resources, check out David Heckerman, Nir Friedman, Dana Pe’er, Eran Segal, Marco Ramoni, Andrew Moore, Jose Pena, and Daphne Koller, Richard Neapolitan, and of course the seminal text by Judea Pearl. All of those are who I am learning from.

Talk 1 (ppt) (horrible introduction, I didn’t know anything)

Talk 2 (ppt) (getting there, decent probability discussion)

Talk 3 (ppt) (getting better, decent graph discussion)

Talk 4 (pdf) (not so great, mostly review, notes only, no slides used)