Basically the idea of providing feedback to achieve a desired state (as instanced by behaviour). When I talk about 'back propagation' I am talking about how these concepts operate at a network level. The important thing to remember about reward and punishment is that it is only one way of inducing a desired state (two others are simple association and personal reflection) which means that the simple adages involving reward and punishment you may hear are frequently contradicted by actual behaviour.
Today: 2 Total: 105 [Direct link] [Share]
