Boltzmann Machines

• A model that is widely used in machine learning is the Boltzmann Machine.
• They use it as a stochastic integrator. That is it doesn't tend to save it's energy from step to step.
• It's outputs are either true or false, so it is quite like a perceptron.
• However, it has a stochastic output. The more energy it has, the more likely it is to fire.
• If this behaviour is combined with a leaky integrate and fire model, the whole mechanism can be used as a time varying neural model.
• This is immediately stochastic.
• Other thresholding models (e.g. adaptation, IF) can also use this method.