Neuroinformatics, 2015-12-01

Module: 
Neuroinformatics
Examiner: 
Gordon Pipa
Assessor: 
Olivera Stojanovic
Date: 
Tue, 2015-12-01

Neuroinformatics / Machine Learning*
Grade: 1.3

Preparation

I studied using almost only the lecture slides.
In hindsight, this was probably not the best way to prepare. Reading the relevant chapters in "Pattern Recognition and Machine Learning" by Christopher M. Bishop is probably a good idea.

Exam

Questions**

Neuroinformatics -- 25 min
  • Probability Theory
  • -What is Bayes Theorem?
    -Can you assign some meaning to the different parts of the equation? (posterior, likelihood, prior)
    -If we wanted to express P(B) differntly, how could we do that? (marginalizing)

  • Model Fitting
  • -What is the idea of parameter estimation?
    What are the steps to get to the solution? why can we take the product?
    -Explain a little bit what the general linear model is.
    What makes it linear? What does that mean for us?
    -What kind of basis functions can we use? What are their properties?

  • Model Selection
  • we talked about various things like overfitting, cross validation, regularization, AIC.

Machine Learning -- 5-10 min
I presented a little bit about clustering. What its purpose is, what different methods there are (hierarchical clustering, k-means, etc), what kind of distance measures you can use and how they affect the clustering.

General remarks

I was given a sheet of paper for notes, graph sketches, etc., but was never asked to derive or calculate anything mathematically, only to verbally outline the basic mathematical ideas, for example in parameter estimation.
All questions were designed to see if I had understood the general concepts presented in the lecture.

Should you plan to take the exam, try to show as much as you can of what you know in the limited time of the exam. Don't hesitate to ask if you don't understand what he means or maybe need a hint.

Good luck :)

-----------------------------------------------------
-----------------------------------------------------

*to be examined in machine learning additionally to neuroinformatics was my choice and not a requirement.
the topic from machine learning which i presented was also determined by me

**the list of questions is incomplete and only serves to give an impression of how questions in the exam could look like