Loading...

Developing a Model for Learning in Spiking Neural Network Domain based on Unique Processing Operator (with Joint Capability of Spatiotemporal Coding)

Iranmehr, Ensieh | 2020

386 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 52693 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Bagheri Shouraki, Saeed
  7. Abstract:
  8. Scientists have discovered by researching the human brain that full knowledge of the functions of the human brain is beyond what they have ever imagined. Much has been learned about the brain and structure of the human nervous system today, but complete structural implementation with the complexity of the human brain is not possible with today's information and technology.One of the biggest struggles while working with artificial neural networks is being able to come up with models which closely match biological observations. Biological neural networks seem to capable of creating and pruning dendritic spines, leading to synapses being changed, which may result in higher learning capability. The latter forms the basis of the present study in which a new ionic model for reservoir-like networks, consisting of spiking neurons, is introduced. High neuroplasticity of this model makes learning possible with a fewer number of neurons. In order to study the effect of the applied stimulus in an ionic liquid space through time, a diffusion operator is used which somehow compensates for the separation between spatial and temporal coding in spiking neural networks and therefore, makes the mentioned model suitable for spatiotemporal patterns. In this proposed reservoir, it is possible to move the dendrites and change the reservoir network topology. For this purpose, we first introduce the reservoir network topology optimization algorithm and then the local learning algorithm for moving the dendrites to enhance the proposed reservoir. The optimization method of the proposed reservoir network topology is based on the genetic algorithm and the local learning algorithm in the proposed reservoir is based on the movement of the dendrites in the opposite direction of ionic gradient. In order to evaluate the proposed reservoir, the separation, approximation and generalization properties are examined and it is observed that the proposed reservoir’s properties are better than that of the original reservoir network (LSM). Moreover, several datasets have been used to evaluate the performance of the proposed model compared to the original LSM. Classification results via separation and accuracy values have shown that the proposed ionic liquid outperforms the original LSM. We have also observed that the topology of the proposed reservoir network can be optimized using genetic algorithm and the proposed local learning algorithm for each problem so as to enhance the separation capability and the classification accuracy
  9. Keywords:
  10. Intrinsic Plasticity ; Evolutionary Algorithm ; Dendritic Plasticity ; Ionic Liquid Space ; Reservoir Networks ; Spiking Neurons ; Structural Plasticity ; Neuroplasticity

 Digital Object List

 Bookmark

No TOC