Reservoir computers were very popular in the early 2000s. From what I understand, the advantage of reservoir computers is that, as opposed to generic recurrent neural networks, training is only done on the linear output layer, making them much faster to train. However, computers are a lot faster today than they were 20 years ago, so I am wondering whether there are still places where reservoir computers are preferred to recurrent neural networks or transformers.
Asked
Active
Viewed 64 times
