NEUROMORPHIC COMPUTING AND INTEGRATIVE COGNITIVE SYSTEMS

iten
Code
101718
ACADEMIC YEAR
2021/2022
CREDITS
6 credits during the 2nd year of 11159 BIOENGINEERING (LM-21) GENOVA
SCIENTIFIC DISCIPLINARY SECTOR
INF/01
LANGUAGE
English
TEACHING LOCATION
GENOVA (BIOENGINEERING)
semester
1° Semester
Teaching materials

OVERVIEW

Several real-world applications already rely upon modules based on neural paradigms as the computational engines of AI algorithms (e.g., deep/convolutional neural networks). The course aims at overcoming the formal framework of artificial neural networks and to relate more decisively to models derived from Neurosciences. A multidisciplinary approach, which links bidirectionally with the Brain Sciences is crucial: from one side, it fosters the transfer towards artificial systems of the knowledge gained from the study of biological systems (i.e., models specified in hardware, software and eventually in wetware that embody in an essential form their principles, architectures and functionalities), and, from the other side, it demonstrates the usefulness of the “artificial” approach as a method for the investigation of the nervous systems.

AIMS AND CONTENT

LEARNING OUTCOMES

Neuromorphic models for the representation and distributed processing of multidimensional signals. Computational primitives and architectural schemes. Applications to the development of perceptual engines to enable autonomous behavior in complex systems and in natural environments.

AIMS AND LEARNING OUTCOMES

Through a reverse engineering of the brain, the course aims at presenting and analysing the basic computational paradigms of cortical processing. Specific emphasis is given to how information from the external world is coded, represented and eventually transformed in the cerebral cortex at network level. The neuromorphic solutions of perceptual problems that support visually-guided behaviour are taken as application domain examples and case-studies.

PREREQUISITES

Foundations of neurosensory systems.

Linear algebra and analytical geometry in space.

Elements of signal processing.

TEACHING METHODS

Traditional lectures will be supplemented by Journal Club and thematic talks on on-going lab activity (48h). Lab practicals on Spiking Neural Network based vision will be offered on a voluntary basis (limited number of participants).

SYLLABUS/CONTENT

Part I 

  • Computational resources
    • Building blocks for simplified functional neural models 
  • Computational paradigms
    • Spatial and spatiotemporal receptive fields in the retinocortical pathway 
    • Regularization and filtering "lessons from computational theory"
    • Multichannel full harmonic representations "lessons from computational theory"
    • Population coding 
  • Architectural paradigms
    • Architectural principles.
    • Lateral inhibition (feed-forward and recurrent, steady-state and dynamic analysis).
    • WTA. Selective amplification. 

Part II - Neuromorphic perceptual engines

  • Foundations on visual motion processing
  • Neuromorphic motion detectors
  • Foundations on stereo-based depth perception
  • Neuromorphc binocular disparity detectors
  • Motion-in-depth

 

RECOMMENDED READING/BIBLIOGRAPHY

Slides and other distributed material (available through Aulaweb).

Bibliography
P. Dayan and L.F. Abbott. Theoretical Neuroscience: Computational and Mathematical Modeling of Neural Systems. The MIT Press, 2001.
H.A. Mallot. Computational Vision: Information Processing in Perception and Visual Behavior. The MIT Press, 2000.

TEACHERS AND EXAM BOARD

Office hours: By e-mail appointment. Office: pad. E, Via Opera Pia 13 (3rd floor) Lab: “Bioengineering ”, pad. E, Via Opera Pia 13 (1st floor)

LESSONS

TEACHING METHODS

Traditional lectures will be supplemented by Journal Club and thematic talks on on-going lab activity (48h). Lab practicals on Spiking Neural Network based vision will be offered on a voluntary basis (limited number of participants).

EXAMS

EXAM DESCRIPTION

Oral examination, evaluation of the Journal Club and continuous assessment of active participation during classes.

ASSESSMENT METHODS

The student should eventually demonstrate:

  • Capacity of formulating a neuromorphic solution of a perceptual problem.
  • Mastery of methods and techniques for a computational solution to an artificial vision problem.
  • Experience of possible application domains.
  • Capacity of understanding a scientific work on the course's related topics.

The oral discussion is aimed at (1) assessing the level of knowledge on the key concepts presented in the course, and (2) verifying the ability to frame and critically analyze the covered topics.

In general, in addition to the correctness and completeness of the answer, the evaluation criteria comprise: the relevance to the question, the clarity of the answer, and the ability to synthesise.

Journal Club operating methods

  • A 20-minute presentation should be prepared in which to summarize the key message of the paper. When the paper is particularly complex and articulated, it will be important to analyze ONLY those aspects that, in your opinion, find more correspondence with the topics covered during classes.
  • To facilitate communication, you are warmly invited to conform the notations of the article to those used in classes.
  • The presentation (typically jointly given by a couple of students) will be followed by a short open discussion in which all are invited to participate.
  • The discussion will be an opportunity to introduce new concepts that will constitute subject of the oral examination
  • The presentation and active participation in the joint discussion will weight, overall, for 50% on the final grade.

Exam schedule

Date Time Location Type Notes
10/02/2022 09:00 GENOVA Esame su appuntamento
15/09/2022 09:00 GENOVA Esame su appuntamento