Difference between revisions of "CSSS 2009 Santa Fe-Tutorials"
From Santa Fe Institute Events Wiki
Line 210: | Line 210: | ||
=== introduction to machine learning and Bayesian inference === | === introduction to machine learning and Bayesian inference === | ||
UPDATE: [[Media:Basic-concepts.pdf | My slides]], with a few improvements. -[[Gustavo Lacerda]] | |||
[[Gustavo Lacerda]]: I'm happy to give a tutorial on basic concepts in statistical machine learning / Bayesian inference (does not touch on reinforcement learning) | [[Gustavo Lacerda]]: I'm happy to give a tutorial on basic concepts in statistical machine learning / Bayesian inference (does not touch on reinforcement learning) | ||
* [[Steven Lade]] I'd like to know more about Bayesian inference | * [[Steven Lade]] I'd like to know more about Bayesian inference |
Latest revision as of 05:56, 30 June 2009
CSSS Santa Fe 2009 |
Feel free to organize your own tutorials.
Post tutorial schedules here.
For ease of scheduling, please make amendments to the schedule
Also, a tutorial outline page wouldn't be bad.
Kolmogorov Complexity
Liliana Salvador: Hey all! Following the Information Theory tutorial, we will give a tutorial on Kolmogorov Complexity and its applications. We will cover the following topics:
- Definition of Kolmogorov Complexity
- Randomness/Incompressibility
- Shannon Information and Kolmogorov Complexity
- Symmetry of Information
- Time Bounded Kolmogorov Complexity
Slides from the talk: Introduction_to_Kolmogorov_complexity.pdf.
Information theory
Discussions and resources are archived at the CSSS 2009 Santa Fe-Information theory tutorial page. Also see this page for recent/current follow up discussion...
MATLAB / Mathematica
Steven Lade: Hiro, I will happily give a tutorial on the basics of these languages (but only the basics, I don't use any of the fancy bits). Anyone here know if the labs here run MATLAB/Mathematica?
If there is anyone else interested (leave your name) I'll arrange a time, otherwise I'll talk directly with you Hiro.
Karen Simpson: I would also like a tutorial on MATLAB. I know some things about it, but never really learned the basics so it takes me a long time to do things. The computer labs should be equipped with MATLAB. I also have a fairly updated version on my laptop.
Lucas LacasaI'm also interested! Also know some basics but I usually program in Fortran so I'd love to learn it.
Steven, many thanks! I think that the computers in the lab have MATLAB. If not, we can access it through my school's server as long as the internet connection is stable. Hirotoshi Yoshioka/lakiaypayaska
Brian Hollar I'm also very interested! I've never used these languages, but have some basic knowledge of Java, NetLogo, and FORTRAN. I'd appreciate the help and would love to learn.
Milena Tsvetkova Count me in! Steve, what is a good time for you?
Wei Ni: Hi ppl. Could I join in? I know MATLAB and I would like to further polish my MATLAB skills. Meanwhile I want to learn Mathematica.
Thanks for coordinating the tutorial and for 'setting up' the facilities, Steve and Lucas.
Chang Yu:Hello,I'm here! Thank you guys for this tutorial.I'm a kid in MATLAB and very courious about it.I know M is a giant in Mathematics,Statistic Analysis,Images and Genetic Algorithm and really want to learn more about that.By the way, I have the installation of Matlab 7.0 and I'll take it in my flash drive.If you have updated version, that would be better.
Steven Lade: Too late for this week I think, and next week is pretty full, so let's aim for 7PM Monday June 22.
Elliot Martin: I would be happy to help out with the MATLAB tutorial if you want. I don't use any of the fancy packages either though.
Marek Kwiatkowski: Excellent tutorial idea, I'll be there.
Margreth Keiler: I would like to join. I've never used these languages but I would like to get an idea.
Liliana Salvador: Good idea. I am in! I have some knowledge of matlab (but never used very fancy packages). I would like to learn Mathematica.
Steven Lade: Here's the outline for the tutorial. I'll introduce basic syntax of MATLAB and code up a basic computational program. Elliot will discuss some visualisation capabilities of MATLAB. Nathan will do some Mathematica basics.
Statistical physics: applications to complex systems
UPDATE: I've uploaded the tutorial slides in Statistical physics: applications to complex systems
Lucas Lacasa:Statistical physics is a rather huge field, so I'm thinking on building a tutorial that focus on some specific topics related to complexity science in a chat-like level, namely:
- Fundamentals of statistical mechanics: ensembles, partition function and associated thermodynamic quantities (free energy, entropy) and some other basic stuff.
- Critical phenomena: Phase transitions in physical, social, and algorithmic systems. Self-organized criticality as the counterpart of a critical phase transition. Relation between phase transitions and local bifurcations of dynamical systems.
- Monte Carlo simulations, ergodic theorem
- Specific example gathering all of the above: Ising model
- Other related topics that you may like to listen to
Maybe we could schedule it for next week, something like next wednesday (17) at 7pm? (provided that no PRIORITY things such as basketball or soccer games are scheduled). If anyone else is interested please leave your name. Depending on the 'audience' we can fix one place or another...
- Steven Lade Fantastic, I'm in. But there is the 'Music on the Hill' 6-8pm Wednesdays. Is 8pm getting too late for 'work'? Or I miss an hour of the music. (Tuesday and thursdays there's the nonlinear dynamics labs)
- Lucas Lacasa You're right. What about friday 19 after lunch?
- Roozbeh Daneshvar I'd like to join this one.
- Corinne Teeter I'd also like to come.
- Barbara Bauer I'm interested too.
- Gustavo Lacerda: I'm interested.
- Allison Shaw: I'd like to join as well.
- watson: maybe you guys should rope
- Andrew Berdahl into this as well. sounds like he could have something to add...
- Elliot Martin: I could talk a bit about self organized criticality, and a bit about some mean field theory if you want some company
- Jacopo Tagliabue: I'm interested, especially if tutorial includes a for-dummies part.
- Massimo Mastrangeli: I will join you!
- Lara Danilova-Burdess: Excellent. I'd like to come.
- Mareen Hofmann: Super! I'm in as well.
- Margreth Keiler: I'm interested, and join the for-dummies part ;-)
- Liliana Salvador: I am in!
- Varsha Kulkarni: I'd like to come!
- Hamid Benbrahim: I would like to come and learn!
Mathematical modeling in ecology
If we want to do this before the talks next week, Monday evening?
- Barbara Bauer whoah, I wish i had an overview on the whole field...
But a subset of this are food webs: recommended readings for the next week's talks of Jennifer Dunne & Neo Martinez are Berlow et al. 2009 Simple predictions of interaction strengths in complex food webs and Dunne et al. 2008 Compilation and network analysis of Cambrian food webs (i can send both if you're interested). This is the latest and probably best of their research agenda, so we could discuss them, and Sasha & me could talk about different approaches/remaining big unsolved questions of community ecology. It would be cool if it'd happen before the talks (this would enable better questions and discussion during the talks), but looking at our schedule, there's not much chance for this...
About other ecol.modeling: I guess I could give a general introduction to it, which will probably be biased towards the stuff i actually know something about- ODE-based modeling, networks and probably adaptive dynamics. For the sake of complexity, I can tell you on what fields they use other types of models (e.g. individual-based or spatial distribution models), but i don't know too much about the details of these.
- Caroline Farrior I agree an overview of the whole field is difficult. I would be willing to take part in this discussion and talk about the kinds of work I do with forest models and models I have seen done in many other areas of Ecology. There a lot of different kinds of models in Ecology which all have their own goals and utilities in combination with data and ideas. I would be very happy to listen to and perhaps contribute to such a discussion.
- Roozbeh Daneshvar I'd like to join. I do not have much background in this area and any amount of review before the relevant classes is highly appreciated.
- Marek Kwiatkowski I would be interested in learning about this too.
- Andrew Noble This all sounds great. I'd be happy to lead a brief overview of stochastic process models in community ecology (focusing on Hubbell's neutral theory of biodiversity). How about next Wednesday at 7pm? I'd like to attend the MaxEnt workshop Monday evening, and there are conflicts with Hubler's nonlinear workshops on Tuesday and Thursday.
- Jacopo Tagliabue I'd like to join, since it sounds very interesting. I do not have much background, so some introductory remarks will be appreciated!
- Lucas Lacasa I'm in
- Steven Lade Yep sounds great -- for scheduling just be aware there's also the 'music on the hill' Wed 6-8pm.
- Barbara Bauer Yeah, you have to choose between music and science here, we have such a tight schedule :( So, the final plan looks like: i will give an outline of math. modeling in ecology in 10 minutes, and after that every ecologist who wants and is there (including me) talks about his-her 'specialty' in cca 5 minutes. Speakers please: forget the details, concentrate on what we already know on your field-what we learned from using your method, and where are the major questions. thx for every contribution!
Game Theory II
Hey All,
I wanted to see if anyone would like a second tutorial on Game Theory. I would assume prior knowledge of simple static (i.e. simultaneous move) games, like the Prisoner's Dilemma and Hawk-Dove game. Rather than focus on these toy models, which are rarely used in practice, I will discuss dynamic (i.e. sequential move) games, with both complete and incomplete information. These models are the most common today and extremely helpful in explaining the logic behind game theory. Topics covered will include:
Modeling Uncertainty
Subgame Perfect Nash Equilibria
Sequential Rationality
Bayesian Nash Equilibria
Advanced Topics: Moving Beyond Bayes
Additionally, if anyone is interested, I can discuss iterated games, stationarity and the Folk Theorem, which provide the foundation for Axelrod's seminal work on the evolution of cooperation. If there is sufficient demand (I'll know it when I see it), I'd be happy to do it early next week (pre-scheduled events permitting, of course), so as to offer a foundation to the coming onslaught of economics, sociology and political science. Trevor Johnston
- Milena Tsvetkova Yes, please! I'm interested in all of the above.
- Caroline Farrior Sounds great.
- Allison Shaw: I'm in.
- Jacopo Tagliabue: I'm in. Yalla Yalla
- Roozbeh Daneshvar: I'm in. As Jacopo said, Yalla!
- Jeremy Barofsky: I am interested and happy to help out. This was one of the best written tutorial proposals I have ever read.
(Evolutionary) Game Theory
Allison Shaw: Since there are a couple requests for this I'd be happy to organize a chat about game theory and evolutionary game theory. I can give an intro to some basic ideas in evolutionary game theory, namely the hawk-dove game, prisoner's dilemma, pure vs mixed vs conditional strategies, Nash equilibrium, evolutionarily stable strategies, and neighbor-invading and convergent stable strategies. Presumably there are people out there with more expertise in general game theory, who could join in...
In terms of a time, since the evenings look pretty crowded, how about at 4:45 after the last lectures, either Tues or Thurs this week, or some day next week?
Post if you're interested.
I'm interested. Marek Kwiatkowski
Lucas Lacasa I'm interested 2
Thanks! Interested as well. Mauricio Gonzalez-Forero
Mareen Hofmann I'd like to join as well! Regarding a time: I'd suggest Thursday after the last lecture.
Chang Yu: Thank u guys. I'm in!
Allison Shaw: Great, let's do this Thursday after the last lecture (4:45ish). I'm just planning to chat about the concepts I listed above, so anyone with more knowledge in game theory is more than welcome to join in.
Hirotoshi Yoshioka: I'm interested too.
Nathan Hodas: Please count me in!
Massimo Mastrangeli: I will join, too.
Roozbeh Daneshvar: I am interested. Liliana Salvador: I will be there.
Allison Shaw: A few people have asked me about recommended books for learning more about evolutionary game theory.
- The only one I am decently familiar with is Evolutionary Games and Population Dynamics, by Josef Hofbauer and Karl Sigmund -- it's fairly comprehensive and is definitely on the mathier end of the spectrum.
- Here's another book I've heard is good, but haven't used: Evolution and the Theory of Games by John Maynard Smith
Machine Learning
reinforcement learning
Hamid Benbrahim: I will be happy to give a tutorial on reinforcement learning if there is intetest. Others with additional expertise are always welcome.
- Milena Tsvetkova I'd be very interested in this tutorial.
- Jacopo Tagliabue I'd be interested too.
- Murad Mithani Me too.
- Roozbeh Daneshvar Me too.
- Sasha Count me in.
Hamid Benbrahim: I scheduled the tutorial Wednesday, July 1st at 1:30PM at SFI
introduction to machine learning and Bayesian inference
UPDATE: My slides, with a few improvements. -Gustavo Lacerda
Gustavo Lacerda: I'm happy to give a tutorial on basic concepts in statistical machine learning / Bayesian inference (does not touch on reinforcement learning)
- Steven Lade I'd like to know more about Bayesian inference
- Wei Ni I am interested in Gustavo's machine learning tutorial proposal.
- Jacopo Tagliabue I am interested.
- Murad Mithani I am in.
- Roozbeh Daneshvar I am in.
Tutorial requests!
Either a formal 'lecture' or a casual 'chat' is fine!
- Time series analysis (requested by Steve Lade)
- I second this request, particularly nonlinear TS and/or using TISEAN (Matt McMahon)
- I would love to hear more about this, especially with respect to Granger causality and VAR. Sasha
- Maximum entropy / Maximum entropy production (requested by Steve Lade)
- Mathematical modelling in ecology (requested by Steve Lade)
- Some physics stuff: statistical mechanics; mean field theory; self-organized criticality and phase transitions; Ising model and the like (requested by Mareen Hofmann and Roozbeh Daneshvar)
- Lucas Lacasa I can talk you about phase transitions, Ising model, SOC and general stat phys...
- Steven Lade I'd like to hear about this too; can you schedule a tutorial?
- Evolutionary game theory (requested by Mareen Hofmann, Roozbeh Daneshvar, Massimo Mastrangeli)
- Ergodic theory (requested by Roozbeh Daneshvar)
- Information theory (requested by Roozbeh Daneshvar and Lucas Lacasa)
- Spectral graph theory (requested by Lucas Lacasa)
- watson, June 16: So Peter Dodds and previously Scott Pauls touched upon this (actually Peter kind of skipped it), but I missed most of that. I would like to either sit in on a tutorial or do a working group...
- Spin glass theory: Replica method (requested by Lucas Lacasa)
- If somebody can give a tutorial on Matlab and/or Mathematica, that would be nice. I'm also interested in the difference between the two programs (e.g., what each program is good at). Thanks in advance! Hiro/lakiaypayaska
- Fitting high dimensional data with functions--is there such thing as 4D maximum likelihood estimation (MLE)? Also how to tell if two 'blobs' of high dimensional data are statistically the same or different.Corinne Teeter
- The MLE is defined for any parameter vector, i.e. any number of dimensions. Finding it is a different story, and can be hard if the objective is not convex. The fully non-parametric ("assumption-free") way to do this is a two-sample test. For 1D data, you could use something like the Kolmogorov-Smirnov test or Cramer-von Mises criterion. Essentially you are stuffing the points from sample 2 into bins defined by the points from sample 1 (or vice-versa): the null hypothesis is that the bins are filled uniformly. You want to find a generalization of this into higher dimensions. I imagine that textbooks on spatial statistics would have this information.
- UPDATE: I think you could use Voronoi cells as your bins, and this works in any number of dimensions! -Gustavo Lacerda
- I propose investigating this topic empirically with a Smirnov test in the upper common room soon. watson
- Bayesian inference. Maximum likelihood estimation. (requested by Steve Lade)
- I'd be happy to give a "casual chat" on this. Gustavo Lacerda
- I'd like to learn about normal modes and the Lagrangian and/or Hamiltonian formulation of classical mechanics. -Gustavo Lacerda
- I am interested in this as well. -Massimo Mastrangeli
- I would be glad to discuss this with you. You might also be interested in how these deterministic systems give rise to randomness Nathan Hodas
- I'd like to learn some beyond-the-basic econometrics and data analysis, possibly focusing on how to build models in social sciences. Anyone? -Jacopo Tagliabue and Roozbeh Daneshvar
- I'd love to get introduced to game theory (Massimo Mastrangeli and Roozbeh Daneshvar and Nathan Hodas)
- Anyone can tell me about graph grammars? (Massimo Mastrangeli)
- I'd second this. The first two links on this search look good, but I know nothing about it... watson
- Could someone give a quick overview of percolation theory? Allison Shaw
- Would anyone be willing to give a primer on artificial intelligence and machine learning? Nathan Hodas
- Note: this is an extremely broad swath of topics you are inquiring about. My advice is to narrow down what you are looking for by at least reading a few sentences from summaries of the sub topics when you search for those key words on Google... watson
- I'll be very interested too in basic topics in machine learning Jacopo Tagliabue
- As Watson said, this is a huge field. I can give you a draft PDF of a new machine learning book (to be published in 2010), which I found very readable. I'd be happy to chat and answer any questions. Gustavo Lacerda
- I will be happy to give a tutorial on machine learning. Hamid Benbrahim
- I understand some of you might be interested in a primer in music? I am available to give away elements of music (from acoustics to notation/rythm to classical harmony (+ elements of composition; no history of music, sorry!). We might even schedule something after 7pm someday in the dining all (we have a piano there for amazing pedagogic purposes). Distributed or walking chats are also possible. -Massimo Mastrangeli
- I am interested in your tutorial on music theory. Hope we can find sometime for it next week. Ciao~ Wei Ni
- Roozbeh: I will join!
- Well, I saw there is a keyboard in second floor common room, we could use that to make examples and illustrate theory. I think the structure of the tutorial may well be interactive, I will provide a tentative set of topics but you can guide the evolution of the tutorial. Tom, David Jones and the other musicians in our group might also help me, if they (as I hope) like to. -Massimo Mastrangeli