Course information, scope and learning outcomes

Course responsible and examiner

Pawel Herman <paherman @ kth.se> 

Teacher

Pawel Herman <paherman @ kth.se>

Registration

If you want to attend the course, you should contact your student office (kansli/studievägledning) of your study programme. There is no maximum of participants, but note the course prerequisits listed in the course and program directory.

Course website

General information about the course can be found at KTH Social Links to an external site.(DD2437).

Course outline

The course consists of 12 lecture themes (including the last summary, Q&A session) and 5 mandatory lab assignments combined with 3 mandatory theoretical quizzes. A list of lectures and labs can be found here in Canvas. 

General course description 

The course is concerned with computational problems in massively parallel artificial neural network (ANN) architectures, which rely on distributed simple computational nodes and robust learning algorithms that iteratively adjust the connections between the nodes heavily using the available data samples. The learning rule and network architecture determine specific computational properties of the ANN. The course offers an opportunity to develop the conceptual and theoretical understanding of computational capabilities of ANNs starting from simpler systems and progressively studying more advanced architectures, and hence exploring the breadth of learning types – from strictly supervised to purely explorative unsupervised mode. The course content therefore includes among others multi-layer perceptrons (MLPs), self-organising maps (SOMs), Boltzmann machines, Hopfield networks and state-of-the-art deep neural networks (DNNs) along with the corresponding learning algorithms. An important objective of the course is for the students to gain practical experience of selecting, developing, applying and validating suitable networks and algorithms to effectively address a broad class of regression, classification, temporal prediction, data modelling, explorative data analytics or clustering problems. Finally, the course provides revealing insights into the principles of generalisation capabilities of ANNs, which underlie their predictive power.

Learning outcomes

After completing the course the student should be able to

  1. Describe the structure and function of the most common artificial neural network (ANN) types, e.g. multi-layer perceptron, recurrent network, self-organizing map, Boltzmann machine, deep belief network, autoencoder, and provide examples of their applications.
  2. Explain mechanisms (algorithms) of supervised and unsupervised learning from data and information processing in different ANN architectures.
  3. Quantitatively analyse the process and outcomes of learning in ANNs, and account for their shortcomings, limitations.
  4. Solve typical small problems in the realm of regression, pattern recognition, scheduling and optimisation with the use of suggested types of ANNs.
  5. (a) Design/devise and (b) implement and optimise ANN approaches to selected real-world problems in pattern recognition, system identification or predictive analytics using commonly available development tools, and (c) critically examine their applicability.

The purpose is to 

  • obtain an understanding of the technical potential as well as advantages and limitations of adaptive, learning and self-organizing systems
  • acquire the ANN practitioner’s competence to apply and develop ANN based solutions to data analytics problems.