The concept of running a neural network with previous input data, for use in the computations of current data. Often seen in RNNs with Recurrent Layer.
Excessive unrolling leads to the Vanishing and Exploding Gradient Problem
The concept of running a neural network with previous input data, for use in the computations of current data. Often seen in RNNs with Recurrent Layer.
Excessive unrolling leads to the Vanishing and Exploding Gradient Problem