mobeets

The following questions can be asked of people (e.g., family members) who somehow manage to form simple opinions about very complex events.

Q1. Where have you learned about the event? First-hand? Or second-hand, through the media?

Q2. Have you ever sought out news about the event?

For example, have you ever had a question about the event and then sought an answer yourself, or do you learn about it as presented via friends and/or the media?

Q3. Have you tried your best to consider the perspective of both sides involved? If so, is there a benefit to you personally in arriving at a definitive opinion?

mobeets

Programs

tl;dr: 1) apply to PhD in: stats/neuro, 2) stats; 3) apply to CNBC program

PhD in Statistics and Neural Computation: This degree is aimed at students who want to join the PNC program, while also specializing in statistics.

Ph.D. Program in Neural Computation: a graduate training program in computational neuroscience for students seeking training in the application of quantitative approaches to the study of the brain.

CNBC Grad Training Program: to be admitted to the CNBC Training Program, students must have been accepted into one of the affiliated doctoral programs at either the University of Pittsburgh or Carnegie Mellon. For students applying from outside the two universities, applications to the doctoral program and the CNBC program may be submitted simultaneously. n.b. not a degree-granting program but I still need to apply

mobeets

Workflow

1. Command line jrn creates file (from template if desired) for specified day, date range, file format
2. Daily builds
   • inputs: *.rtf, *.txt, *.md, *.html, *.png. *.jpeg
   • output: *.md
   • dirs: Journal/ -> _posts/
   • update _data/counts.json
3. Jekyll template for viewing/searching posts

mobeets

Paper: Markov Chain Monte Carlo Methods for State-Space Models with Point Process Observations

Model

• P(Y_k | X_k, mu, beta) ~ Poisson(lambda_k)
• lambda_k = exp(mu + beta*X_k)
• x_k = px_{k-1} + alphaI_k + e_k

where e_k normally distributed with zero mean and known variance; also, I, alpha, and beta are known.

mobeets

Data

• Wyeth Bair's Neural Signal Archive
• Neurodatabase

mobeets

Gibbs sampling to estimate AR(1) model parameters.

Source: SSC 389: AR(1) + noise

Definitions

• Xt is hidden state at time t
• Yt is observed state at time t
• Dt is information known at time t, i.e. {Y1, ..., Yt}
• T is length of X, Y, D

mobeets

Definitions

SNR = Power(signal) / Power(noise)

Power(x(t)) = average Energy(x(t)) over all t

Energy(x(t)) = the area under |x(t)|^2

(Energy/power involve squaring the signal because you are interested in the magnitude of the signal and not just its amplitude. That's it.)

mobeets

Defintions

For some matrix A:

• A is symmetric ⇔ A = A'
• A is positive definite ⇔ for every non-zero vector x, x'Ax > 0
• A is non-singular ⇔ A is invertible ⇔ detA≠0 ⇔ there exists B such that AB = BA = I
• A is a distance matrix ⇔ A is symmetric and diag(A) = 0

Fun facts

1. A is non-singular ⇒ AA' is positive definite
2. A is symmetric ⇒ AA is positive definite

mobeets

This was surprisingly simple given how many things there were to do! (Usually it seems like something breaks when there's this many steps between me and installing something.)

Installations

• Install Sublime Text 2
• Install Sublime Text's Package Control
• Configure things so you can open files with subl from command line
• Install SublimeREPL using Pacakge Control.
• Install homebrew
• Install leiningen by typing brew install leiningen at the command line.

mobeets