Improving the performance of spatial searching in BTMORPH: An investigation into the implementation of R-tree spatial indexing

In order to improve BTMORPH’s spatial querying capabilities, an implementation of Guttman’s (1984) R-tree and an investigation to determine the best parameters for performance was conducted. This presentation will discuss:

  • the precursory improvements to work flow
  • the implementation of forest structures that were a necessary to the addition of spatial indexing into BTMORPH
  • the effects on the R-tree of the parameter M
  • the method of node splitting during R-tree construction
  • the BTMORPH forest handling strategy

All of which were used to produce a 400% reduction in average search time on a 130,000 node search.

Date: 10/02/2017
Time: 16:00
Location: LB252

Share this post on: Twitter| Facebook| Google+| Email

Categories

Tags

neuroscience Computer Science COVID-19 PCA Persistent homology Convolutional Neural Networks Deep learning Associative memory Purkinje cell cerebellum olfaction computational neuroscience Neuromorphic hardware Dendritic computation Evolutionary algorithms Software development optimization Open position Structural plasticity neural networks robotics Neurons Studentship Neuronal Morphology Artificial intelligence computational modelling Homoeostasis Network connectivity synaptic plasticity machine learning