Study unit
3621675 Deep Learning, 6 Cp  
Code 3621675  Validity 01.01.1950 -
Name Deep Learning  Abbreviation DEL 
Credits 6 Cp Date of expiry  
TypeAdvanced Studies Subject0530 Computer Science 
ClassStudy Unit  Hours  
Study right   Grading0-5 
Recommended scheduling 
   
Organisation Computer Science (J,K) 

Description
Learning outcomes 

Students understand the underlying principles of machine learning and data mining. Students understand the theory of neural networks, Nature-Inspired Computing (NIC) methods, and deep learning algorithms. Students get familiar with the most popular neural networks models (Perceptron, Adaline, multilayer neural networks, SOM, etc.) and their learning algorithms (Widrow-Hoff rule, Back-Propagation (BP) training, Back-propagation Through Time (BTT) training, competitive learning, etc.). Students can apply learning algorithms for adjusting the weights in these neural networks. Students understand the basics of optimization theory and techniques. Students know the most widely applied swarm algorithms (PSO, ACO, DE, etc.). Students are able to use these swarm intelligence methods for solving optimization problems. Students understand the working principles of a few representative reinforcement learning methods (Adaptive Critic Design (ACD), Temporal Difference (TD), etc.). Students master the essential knowledge of deep learning techniques, such as regularization, momentum, batch normalization, and dropout. Students understand the principles, structures, and algorithms of typical deep learning neural networks, e.g., CNN. Students know how to build up deep learning algorithms from scratch. Students know how to choose evaluation metrics, how to preprocess data sets, and how to handle variance and bias in deep learning. Students are able to program deep learning using the TensorFlow platform. Students gain the hand-on experiences in using deep learning techniques to deal with practical problems.

 
Content 

Elementary concepts and challenges of machine learning. Neural networks models (linear neural networks, feedforward neural networks, recurrent neural networks, Self-Organizing Map (SOM), etc.). Meta-heuristics optimization methods (Genetic Algorithms (GA), Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), Differential Evolution (DE), etc.). Reinforcement learning. Concepts and challenges of deep learning. Deep learning models and techniques (deep neural networks, Convolutional Neural Networks (CNN), etc.). Applications of deep learning in classification, prediction, pattern recognition, etc.

 
Modes of study 

Lectures, teaching materials, exercises, and examination.

 
Teaching methods 

Lectures 32 hours, exercises 16 hours.

 
Study materials 

I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning, MIT Press, 2017, A. P. Engelbrecht, Fundamentals of Computational Swarm Intelligence, John Wiley & Sons Ltd, 2005, S. Haykin, Neural Networks, A Comprehensive Foundation, Prentice Hall, 2008, lecture slides and notes, selected papers from journals and conference proceedings.

 
Evaluation criteria 

Grading: 0-5

 
Prerequisites 

Basic knowledge of linear algebra and probability theory, general MATLAB programming skills

 


Letter (J, K) in front of the name of the course/exam indicates the campus on which teaching or exam takes place: J = Joensuu, K = Kuopio.

Present and future teaching
Functions Name Type Cp Teacher Timetable
Registration ended
(KUOPIO): Deep Learning  Lecture and exercise course  Xiaozhi Gao  29.10.19 -31.01.20
Registration ended
(JOENSUU): Deep Learning  Lecture and exercise course  Xiaozhi Gao  29.10.19 -31.01.20

Future exams
No exams