Virginia Tech® home

ECE 5440 - Brain-inspired Computer Architecture (3C)

Course Description

The structure of biological brains as the only truly intelligent systems that exist. Comparisons between computing elements such as transistors and processing elements and brain structures such as neurons. Complexity of the brain and limitations in conventional computer parallelism. State of the art brain-inspired computer architectures. Novel programming models and architectures as bases for proposing brain-inspired computer architectures at the system level. Foundations for intelligent discussion of how the brain relates to present and future computer architecture. Structural bases for and implications of true machine intelligence.

Why take this course?

Even as Moore’s Law and conventional forms of computer parallelism are hitting limits and abundant research is being done in computational methods of machine learning, the relationship between brain structure and intelligence remains a mystery.

Biological brains are the only intelligent systems that exist. Current research in artificial intelligence and machine learning is focused on creating complex programs that require large datasets for training. Furthermore, it does not provide insight into how the physical structure of the brain achieves intelligence, utilizes massive amounts of parallelism, balances accuracy and real-time response, and creates new solutions from relatively few experiences. In addition, state-of-the-art “continual learning” systems suffer from things like “catastrophic forgetting” suggesting more than mathematical models are required to achieve true machine intelligence.

To achieve true machine intelligence, understand the structure of the brain from the perspective of a computer engineer, or inspire novel computer organization, the structural organization of the brain must be understood in the context of its potential impacts on future computer architecture.

Learning Objectives

  • Compare and contrast neurons with transistors and processing elements.
  • Calculate the complexity and parallelism of the brain and contrast it with current computer architecture.
  • Describe limitations of existing computer architecture in achieving levels of complexity and parallelism in the brain.
  • Analyze the state of the art of current approaches towards implementation of brain-inspired architectures.
  • Propose novel designs of brain-inspired computer architecture.