HLP Lab Meeting Schedule

Spring Semester '13

Wednesdays at 1:30pm

Possible readings and topics

• (Bozena) Alon (2009) - How To Choose a Good Scientific Problem Alon_2009.pdf

• (Alex) hierarchical Bayesian models (what is a chinese restaurant process? what is an indian buffet?). This could (should) stay at a conceptual level, and address questions like
  • what kinds of problems do these approaches lend themselves to?
  • for these problems, what kinds of theoretical or computational approaches were people taking before these hierarchical models existed?
  • similarly, what are the alternatives/objections to these models nowadays?
  • how are hierarchical bayesian models related to/different from multi-level regression?
• (Alex) what do people really mean when they say "learning"? (maybe dick could be consulted about this?)
  • what is the difference between supervised and unsupervised learning?
  • what sub-types of learning exist within these categories?
  • large parts of the field seem to have essentially forgotten about "classical" approaches to learning (associative learning, conditioning, etc.)
• (Alex) Dry run for CUNY talk? maybe me and chigusa on the same day.
• (Bozena) For later in the semester (perhaps mid/late March or so) I would be interested in presenting some modeling work I've been doing.

• (Bozena) If we have a meeting on the general topic of how to get organized, I could talk about the software I'm using, OmniFocus, which implements David Allen's "Getting things done" philosophy. I could talk a bit about the general philosophy, and then show how the software works. It'd probably take about 15-20 minutes

• (Dan) I think tutorials about stats and modeling would be most useful for me. I would also be interested in learning about study design constraints/rules with methodologies that I have yet to use (self paced reading) and definitely corpus analysis tutorials.

• (Chigusa) I would be interested in reading some papers on assumptions about sampling in learning (like Xu and Tenenbaum and also the following paper) SAMPLING ASSUMPTIONS IN LANGUAGE LEARNING Anne Hsu and Thomas Griffiths (submitted)

• (Ting) I've recently come across some nice papers that might be of interest to the lab in general:

  • Beck, J. M., Ma, W. J., Pitkow, X., Latham, P. E., & Pouget, A. (2012). Not noisy, just wrong: the role of suboptimal inference in behavioral variability. Neuron, 74(1), 30–9. doi:10.1016/j.neuron.2012.03.016

  • Lagnado, D. a, & Sloman, S. a. (2006). Time as a guide to cause. Journal of experimental psychology. Learning, memory, and cognition, 32(3), 451–60. doi:10.1037/0278-7393.32.3.451

  • Pellicano, E., & Burr, D. (2012). When the world becomes “too real”: a Bayesian explanation of autistic perception. Trends in cognitive sciences, 16(10), 504–10. doi:10.1016/j.tics.2012.08.009

  • Cohen, J. D., McClure, S. M., & Yu, A. J. (2007). Should I stay or should I go? How the human brain manages the trade-off between exploitation and exploration. Philosophical transactions of the Royal Society of London. Series B, Biological sciences, 362(1481), 933–42. doi:10.1098/rstb.2007.2098

• (Sarah) I would appreciate people talking about how they use python and/or pyjamas for their projects. Not a full programming tutorial because I'm sure we don't have time for that (also I thought Steve Piantadosi was going to continue with his, but I never heard what happened to the second half....). However I would be interested in hearing about in general what is good about python; what is it capable of doing, what kinds of projects is it useful for, what are the downsides to it...etc. And the same kind of thing for pyjamas. It would also be nice to know what kinds of projects people have already programmed and who would be willing to share code. Currently right now I don't have expertise in any of the programming languages people use to set up experiments (exbuilder, matlab, python) and so it would be nice to know which might be the best for me to focus on for now.

• (Masha) Last week, I suggested talking about replicability. There's an issue of PPS on replicability (http://pps.sagepub.com/content/7/6.toc)

• (Ilker) I could do one or two (or zero) tutorials. One on Bayesian nonparametrics. That could contain stuff such as Dirichlet Processes, Indian buffet process. But also could contain nonparametric hierarchical clustering models, which might be of interest to (phonetic) acquisition people (e.g., Dirichlet Diffusion Trees, Neal, 2004; and newer, shinier stuff.). The other on MCMC for structured representations. We know some about vanilla Gibbs sampling, Metropolis Hastings, which I dont mind talking about. But often, we end up doing inference over highly sophisticated structures such as probabilistic grammars, shape grammars, programs, etc. In such cases, MCMC methods such as tree-based MCMC, type-based MCMC come handy. I could talk about these things.
• (Dave) Here are ideas I had for tutorials etc. for lab meetings, ordered roughly from most to least favored:
  • plyr/reshape for manipulating data. I find these really useful for doing data exploration and analysis but it took me a while to figure out how to use them effectively.
  • Sweave/knitr
  • Using a version control system (like git). Maybe with an emphasis on "non-traditional" applications (like managing data and code for experiments)