Properties of the Backpropagation Algo.

• Learns weights for a multilayer network, given a fixed set of units and interconnections
• It uses gradient descent to minimize the squashed error between the network outputs and the target values for these outputs.
• 
• is the set of output units in the network, and are the target and output values associated with the th output unit and the training example
• In multilayer networks the error surface can have multiple minima, but in practice Backpropagation has produced excellent results in many real-world applications
• the algorithm is for two layers of sigmoid units and does stochastic gradient descent