Introduction to Neural Networks
U. of Minnesota, Final Study Guide
Psy 5038

Fall, 2014

The final exam will primarily cover material from the second half of the semester.

Sample short answer questions
Define and describe the relation of the following key words or phrases to neural networks. Provide examples where appropriate.(Answer 8 out of 12 items drawn from set below; 3 points each).

K-means topographic representation "S" vs. "C" units EM

Gibbs sampling Oja's rule principal components analysis sparse distributed representation

directed vs. undirected graph K-nearest neighbor classifier "explaining away" support vector machine

grandmother cell Bayesian optimal cue integration encoder network topology-preserving map (Kohonen)

contingent adaptation cortical maps loss and risk functions belief propagation

MAP estimation MCMC
logistic regression
anti-Hebbian

Bayesian decision theory projective field bidirectional processing coarse coding

marginalization & conditioning "features of intermediate complexity" rejection sampling MRF

Sample essay questions
(choice of 2 essays drawn from those listed below; 12 points each).

Discuss the pros and cons of distributed vs. localized representations with examples from theoretical considerations and neurophysiology (e.g. Quiroga et al., 2005).

Explain how Belief Propagation can be used to interpolate a function given sparse data.

Give an account of just one of the following approaches to self-organization: a) Kohonen, 1982; or b) principal components for dimensionality reduction.

Discuss how probabilistic information might be represented in a population of neurons, and how that information could be used for optimal inference (Pouget et al. 2006; Knill and Pouget, 2004; Ma, 2012).

Describe the Metropolis-Hastings algorithm.

What is a mixture model? How can EM be used to estimate the parameters of the model?

Describe and discuss Serre et al's (2007) feedforward model of recognition. Could it help to explain Kirchner & Thorpe's 2006 result?

Describe and discuss Ullman et al's (2002) feedforward model of recognition. Explain how to extend this to a bidirectional model of visual recognition (Epshtein et al., 2008).

What are possible computational functions of bidirectional processing between cortical areas? (see Bullier , 2001; Kersten & Yuille, 2014)

Explain how the Kalman filter can be used to model neural processing or behavior (Rao & Ballard 1999 or Wolpert et al. 1995).


K-means	topographic representation	"S" vs. "C" units	EM
Gibbs sampling	Oja's rule	principal components analysis	sparse distributed representation
directed vs. undirected graph	K-nearest neighbor classifier	"explaining away"	support vector machine
grandmother cell	Bayesian optimal cue integration	encoder network	topology-preserving map (Kohonen)
contingent adaptation	cortical maps	loss and risk functions	belief propagation
MAP estimation	MCMC	logistic regression	anti-Hebbian
Bayesian decision theory	projective field	bidirectional processing	coarse coding
marginalization & conditioning	"features of intermediate complexity"	rejection sampling	MRF