West Virginia University
Spiking Neural Network that Maps from Generalized Coordinates to Cartesian Coordinates
Abstract
dc:description.abstractIn this thesis, I look to understand how insects compute task-level quantities by integrating range-fractionated sensory signals to create a sparse-spatial coding of Cartesian positions. I created biologically plausible 2-D and 3-D models of one species of the stick insect (Carausius morosus) leg and encoded the foot position through a spiking neural network. This model used spiking afferents from three angles of an insect leg which are integrated by one non-spiking interneuron. This model contains many dendritic compartments and one somatic compartment that encode the foot’s position relative to the body. The Functional Subnetwork Approach (FSA) was used to tune the conductances between the compartments (Szczecinski et al., 2017). Also, the Product of Exponentials (POE) was used to calculate the spatial kinematic chain of the stick insect leg (Murray et al., 1994). The system accurately encodes the foot position and depends on the width of the sensory encoding curves, or the “bell curves”. Discussion of limitations and other studies that relate to this work, as well as motivation for future work are included.
Degree
thesis:*- Name thesis:degree_name
- MS
- Level thesis:degree_level
- Thesis
- Discipline thesis:degree_discipline
- Mechanical and Aerospace Engineering
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Guie, Chloe K
- Contributors dc:contributor
-
- Nicholas Szczecinski
- Sergiy Yakovenko
- Xi Yu
Subjects
dc:subject × 7Identifiers
dc:identifier.*- Identifier
- https://researchrepository.wvu.edu/etd/12216