The purpose of this website is to publicly discuss the TempUnit parallel architecture for artificial intelligence and robotic isssues.
The TempUnit Parallel architecture, will be defined clearly (I hope) through articles on this website.
But if you cannot wait, you can just read/download:
- Temporal and rate decoding in spiking neurons with dendrites
- Local learning rules for spiking neurons with dendrite
- TempUnit: A bio-inspired neural network model for signal processing.
TempUnit is not a fixed on the rock neural network architecture. It can evolve or be modified to fit new neuroscience discoveries. But also, the goal is not only to reproduce all the discoveries made on biological neurons but at the opposite to take only the core biological characteristics that are necessary to solve complex « real life » problems.
The problems to solve are:
- discrimination: detection of patterns in a sensory signal
- motion control: one or more articulated joint chains.
- decision: association between an input and an output.