### Neural Networks

Data is displayed for academic year: 2023./2024.

#### Laboratory exercises

#### Course Description

The course enables students to gain knowledge of the theory and applications of artificial neural networks.

The following topics are covered: Biological and artificial neural networks. Models of neurons. Activation function. Network topologies. Perceptron. Laws of learning. Associative networks. Linear association. Multilayer networks. Delta rule for error back propagation. . Support vectors machines. Radial basis function networks. Recursive networks. Hopfield Network. Energy function. Boltzmann machine. Simulated annealing. K-means algorithm. Kohonen's self-organizing network. Simulation software packages. Applications in pattern recognition and in signal and image analysis.

The following topics are covered: Biological and artificial neural networks. Models of neurons. Activation function. Network topologies. Perceptron. Laws of learning. Associative networks. Linear association. Multilayer networks. Delta rule for error back propagation. . Support vectors machines. Radial basis function networks. Recursive networks. Hopfield Network. Energy function. Boltzmann machine. Simulated annealing. K-means algorithm. Kohonen's self-organizing network. Simulation software packages. Applications in pattern recognition and in signal and image analysis.

#### Study Programmes

##### University graduate

[FER3-EN] Data Science - profile

Elective courses
(1. semester)
Recommended elective courses
(3. semester)
#### Learning Outcomes

- Understanding the basic concepts of neural networks
- Ability to create solutions based on neural networks
- Ability to adapt existing neural networks to a new problem
- Ability to use existing software frameworks for neural networks
- Ability to evaluate the performance of neural network based solutions

#### Forms of Teaching

**Lectures**The lectures present theoretical concepts and algorithms followed by concrete examples.

**Laboratory**Laboratory exercises are done with computers. During the exercise, students try out theoretical concepts and apply them to specific problems

**Other**Team project in which students solve a real practical problem of biomedical image analysis

#### Grading Method

Continuous Assessment | Exam | |||||
---|---|---|---|---|---|---|

Type | Threshold | Percent of Grade | Threshold | Percent of Grade | ||

Laboratory Exercises | 50 % | 20 % | 50 % | 20 % | ||

Seminar/Project | 20 % | 20 % | 20 % | 20 % | ||

Mid Term Exam: Written | 20 % | 30 % | 0 % | |||

Final Exam: Written | 20 % | 30 % | ||||

Exam: Written | 50 % | 60 % |

##### Comment:

The threshold on the sum of the midterm and the final exam is 50%.

#### Week by Week Schedule

- Perceptron (learning paradigms,Hebbian learning, competitive learning, Boltzmann learning)
- Perceptron (learning paradigms,Hebbian learning, competitive learning, Boltzmann learning)
- Perceptron (learning paradigms,Hebbian learning, competitive learning, Boltzmann learning)
- Multilayer perceptron (error-backpropagation learning, credit-assignment problem, backpropagation through time)
- Multilayer perceptron (error-backpropagation learning, credit-assignment problem, backpropagation through time)
- Radial basis function networks (solving interpolation problem with radial basis function networks, generalized radial basis function networks, relation to regularization theory)
- Support vector machine for classification
- Midterm exam
- Recurrent neural networks (Hopfield network, Hopfield network energy function, Boltzman machine, Elman networks, Jordan networks) and learning algorithms (back propagation through time, reccurent backpropagation)
- Self-organizing networks (Hebbian non-supervised learning, Oja's learning rule, PCA using self-organizing network, Sanger's learning rule, Competitive non-supervised learning, winner-takes-all network, Kohonen's self-organizing maps)
- Network ensembles (committee machines, mixture of experts, convolutional neural networks)
- Deep convolutional networks: layers, architectures, visualization, fine tuning, applications, implementation
- Deep generative models: stacked RBMs, convolutional autoencoders, variational autoencoders, adversarial models, sparsity
- Project
- Final exam

#### Literature

Simon S. Haykin (2009.),

*Neural Networks and Learning Machines*, Prentice Hall
James A. Anderson (1995.),

*An Introduction to Neural Networks*, MIT Press
Charu C. Aggarwal (2023.),

*Neural Networks and Deep Learning*, Springer Nature#### For students

#### General

**ID**222982

Winter semester

**5**ECTS

**L1**English Level

**L2**e-Learning

**30**Lectures

**0**Seminar

**0**Exercises

**15**Laboratory exercises

**0**Project laboratory

**0**Physical education excercises

#### Grading System

**87**Excellent

**75**Very Good

**63**Good

**51**Sufficient