Competitive Networks
- I tried to get this stuff working following Diehl and Cook's (2015) paper
but was not very successful.
- I then reread that paper, and saw that it derived from, Habenschuss
et al (2012) Homeostatic plasticity in Bayesian spiking networks
as Expectation Maximization with posterior constraints.
- They use a spiking net with Hebbian plasticity, but the
structure is a winner take all system.
- The inputs connect to the categorisation neurons.
- Each categorisation neuron connects to all of the others,
inhibiting them.
- Using infinite speed synapses, exactly one neuron wins for
any input, and inhibits the others.
- The only plasticity is between the inputs and the categorisation
neurons.
- With mathematical precision, a network can associate particular
categorsiation neurons with particular inputs.
- It's like a self organising map.
