Course detail

Evolutionary and Neural Hardware

FIT-EUDAcad. year: 2020/2021

This course introduces selected computational models and computer systems which have appeared at the intersection of hardware and artificial intelligence in order to address insufficient performance and energy efficiency of conventional computers in solving some hard problems. The course surveys relevant theoretical models, circuit techniques and computational intelligence methods inspired in biology. In particular, the following topics will be discussed: evolutionary design, evolvable hardware, neural hardware, neuroevolution and approximate computing. Typical applications will illustrate these approaches.

Doctoral state exam - topics:

  1. Inspiration in biology (adaptation, self-organization, entropy, evolution, learning).
  2. Hardware and reconfigurable devices for artificial intelligence.
  3. Cartesian genetic programming.
  4. Scalability issues of evolutionary circuit design and their solutions.
  5. Evolutionary design of analog circuits.
  6. Cellular automata in 1D and 2D, Wolfram classes, self-replication.
  7. Approximate computing (principles, error metrics, circuit approximation methods).
  8. Deep neural networks.
  9. Hardware implementation of neural networks.
  10. Neuroevolution.

Language of instruction

Czech

Mode of study

Not applicable.

Learning outcomes of the course unit

Students will be able to utilize evolutionary algorithms to design electronic circuits. They will be able to model, simulate and implement bio-inspired computational systems, particularly evolvable and neural hardware.
Understanding the relation between computers (computing) and some natural processes.

Prerequisites

Not applicable.

Co-requisites

Not applicable.

Planned learning activities and teaching methods

Not applicable.

Assesment methods and criteria linked to learning outcomes

Submission of the project on time, exam.

Course curriculum

Not applicable.

Work placements

Not applicable.

Aims

To understand the principles of bio-inspired computing techniques and their use particularly during the design, hardware implementation and operation of computer systems.

Specification of controlled education, way of implementation and compensation for absences

During the course, it is necessary to submit the project and pass the exam. Teaching is performed as lectures or controlled self-study; the missed classes need to be replaced by self-study.

Recommended optional programme components

Not applicable.

Prerequisites and corequisites

Not applicable.

Basic literature

Not applicable.

Recommended reading

Floreano, D., Mattiussi, C.: Bioinspired Artificial Intelligence: Theories, Methods, and Technologies. The MIT Press, Cambridge 2008, ISBN 978-0-262-06271-8
Reda S., Shafique M.: Approximate Circuits - Methodologies and CAD. Springer Nature, 2019, ISBN 978-3-319-99322-5
Trefzer M., Tyrrell A.M.: Evolvable Hardware - From Practice to Application. Berlin: Springer Verlag, 2015, ISBN 978-3-662-44615-7

Classification of course in study plans

  • Programme CSE-PHD-4 Doctoral

    branch DVI4 , 0 year of study, summer semester, elective

  • Programme CSE-PHD-4 Doctoral

    branch DVI4 , 0 year of study, summer semester, elective

  • Programme CSE-PHD-4 Doctoral

    branch DVI4 , 0 year of study, summer semester, elective

  • Programme CSE-PHD-4 Doctoral

    branch DVI4 , 0 year of study, summer semester, elective

Type of course unit

 

Lecture

26 hod., optionally

Teacher / Lecturer

Syllabus

  1. Introduction.
  2. Bio-inspired computational models (inspiration, principles of adaptation and self-organization).             
  3. Approximate computing and energy efficiency.
  4. Hardware and reconfigurable devices for artificial intelligence.
  5. Evolutionary design.
  6. Cartesian genetic programming.
  7. Evolutionary design of digital and analog circuits.
  8. Scalability problems of evolutionary design.
  9. Computational development, cellular automata, L-systems.
  10. Deep neural networks and their hardware implementation.
  11. Approximate computing for neural networks.
  12. Neuroevolution.
  13. Recent HW/SW platforms and applications.

Guided consultation in combined form of studies

26 hod., optionally

Teacher / Lecturer