Exam problems
1.
Types of neural networks (architecture + dynamics), learning paradigms
2.
Binary perceptron

definition, learning algorithms
(Hebb, Perceptron, Adatron)

heuristic
explanation, proof of convergence of perceptron algorithm
.
3.
Generalization err
or

definition and theoretical calculations for binary perceptron.
4.
Delta and error back propagation algorithms
.
5.
Advanced gradient

based learning algorithms: st
eepest descent, variable metric (quasi
Newton)
,
Levensberg

Marquard
6.
Advanced gradient

based lea
rning algorithms:
conjugate gradients with regularization, choice
of learning coefficients in gradient

based algorithms
7.
Heuristic and global learning algorithms for neural networks
8.
Algorithms for constructing neural networks (OBD, OBS, ...), practical aspe
cts of training
neural network (overfitting, underfitting, early stopping, cross

validation)
9.
Practical applications of multi

layer networks: function approximation, classification, data
compression.
10.
Definition of a recurrent network in the language of stat
istical mechanics: stochastic neuron,
Hamiltonian, order parameters
.
11.
Hopfield model: definition, phase diagram, solution exploiting methods of statistical
mechanics
.
12.
Maximal capacity of a neural network

combinatorial and Gardner approach.
13.
Boltzmann machi
ne: definition, learning algorithm and applications
14.
Unsupervised learning: Hebb rule and autoencoder supported by a Boltzmann machine.
15.
Unsupervised learning: vector quantization.
16.
Unsupervised learning: Kohonen network

definition, motivation and applicati
on.
17.
Radial basis function network.
18.
Support vector machine.
19.
General structure and inspiration of genetic algorithms, action of a simple genetic algorithm.
20.
Schemata theorem and its
criticism
.
21.
Choice of representation and genetic operators.
22.
Selection and sca
ling operators.
23.
Crossover and mutation operators for binary and floating point representation.
24.
Solution of problems with constraints.
25.
Transport task

solution with GA: representation and genetic operators.
26.
Traveling salesman problem

solution with GA: re
presentation and genetic operators.
27.
Practical application
s
of genetic algorithms

3 examples
: medical imaging, robot's route
planning, design of buildings
Comments 0
Log in to post a comment