Restricted Boltzmann Machine-Assisted Estimation of Distribution Algorithm for Complex Problems
Algorithm 1
Pseudocode of the probabilistic modelling.
Begin
Do while (maximum number of training epochs is not reached)
#Positive Phase
1. Construct the conditional probability of the hidden units given the visible softmax values according to (6)
2. From , sample the states of the hidden units
#Negative Phase
3. Construct the conditional probability of the visible softmax units given the states of the hidden units according to (7). Reconstruct the states of the visible softmax units by sampling the constructed conditional probability
4. Construct the conditional probability of the hidden units given the sampled visible softmax values according to (6). Reconstruct again the states of the hidden units by sampling the constructed conditional probability
#Updating of weights
5. Update the weights and biases
End Do
#Construction of the probability model
6. Calculate the Probability Model according to (8).
