Dynamic synapse is a learning rule we proposed recently based on biological findings. The synaptic strength in this model is dynamic, which is different from conventional model whose synaptic strength is static. Here we presents an simplified viersion of the learning rule. For more details, please see our recent paper accepted by IROS 2018 named A Bio-inspired Reinforcement Learning Rule to Optimise Dynamical Neural Networks for Robot Control.
The video of the experiment in above paper can be found following the link:https://youtu.be/B7mLVY1NKgI
There is a more complex and biologically plausiable version of this model. For detials of this version, please see our another paper A model of operant learning based on chaotically varying synaptic strength.