Convergence of learning algorithms with constant learning rates
{{output}}
The behavior of neural network learning algorithms with a small, constant learning rate, epsilon, in stationary, random input environments is investigated. It is rigorously established that the sequence of weight estimates can be approximated by a certain ordi... ...
