Figure 4
From: Using Memristors for Robust Local Learning of Hardware Restricted Boltzmann Machines
(a) Test error rate achieved by the Discriminative RBM as a function of the programming pulse width for different mini-batch sizes and different number of parallel Gibbs chains to evaluate the Contrastive Divergence term (‘# CD’) in the near-linear case (β = 0.005). (b) Same Figure as (a) in the non-linear case (β = 3). (c) Optimal test error rate achieved by the Discriminative RBM for different values of β with different mini-batch sizes and different number of parallel Gibbs chains to evaluate the Contrastive Divergence term (‘# CD’). For mini-batches size of 100, each simulation was ran over 30 epochs, 5 times per value of pulse width, error bars indicate median, first quartile and third quartile. For mini-batches of size 1, each simulation was ran over 50 epochs 5 times per value of pulse width to ensure convergence.
