jens/pyRBM
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

train: sample gaussian visible units

Browse Source
This commit is contained in:
jens
2026-01-03 12:40:08 +01:00
parent 63bfd3ed80
commit 529252166a
1 changed files with 21 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/rbm/train.py
+21 -9
View File
@@ -63,6 +63,7 @@ def cd_jens(entity: Entity, v_states: Mat):
# %%%%%%%%% END OF NEGATIVE PHASE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% # %%%%%%%%% END OF NEGATIVE PHASE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
def cd_binary_binary(entity: Entity, data_pos: Mat): def cd_binary_binary(entity: Entity, data_pos: Mat):
params = entity.training_params params = entity.training_params
# Positive phase # Positive phase
h_probs_pos = prob(entity.h_given_v(data_pos)) h_probs_pos = prob(entity.h_given_v(data_pos))
@@ -92,19 +93,25 @@ def cd_binary_binary(entity: Entity, data_pos: Mat):
return dw, dbv, dbh return dw, dbv, dbh
def cd_gaussian_binary(entity: Entity, data_pos: Mat): def cd_gaussian_binary(entity: Entity, data_pos: Mat):
params = entity.training_params
# Positive phase # Positive phase
h_probs_pos = entity.h_given_v(data_pos) if params.do_batch_sample:
h_probs_pos = entity.h_given_v(sample_gaussian(data_pos))
else:
h_probs_pos = entity.h_given_v(data_pos)
# Update weights (positive phase) # Update weights (positive phase)
dw = np.dot(np.transpose(data_pos), h_probs_pos) dw = np.dot(np.transpose(data_pos), h_probs_pos)
dbh = np.sum(h_probs_pos, 0) dbh = np.sum(h_probs_pos, 0)
dbv = np.sum(data_pos, 0) dbv = np.sum(data_pos, 0)
# Sample hidden states if not params.do_rao_blackwell:
h_states_pos = sample(h_probs_pos) # Sample hidden states
h_probs_pos = sample(h_probs_pos)
# Negative phase # Negative phase
data_neg = entity.v_given_h(h_states_pos) data_neg = entity.v_given_h(h_probs_pos)
h_probs_neg = entity.h_given_v(data_neg) h_probs_neg = entity.h_given_v(data_neg)
# Update weights (negative phase) # Update weights (negative phase)
@@ -115,19 +122,24 @@ def cd_gaussian_binary(entity: Entity, data_pos: Mat):
return dw, dbv, dbh return dw, dbv, dbh
def cd_gaussian_gaussian(entity: Entity, data_pos: Mat): def cd_gaussian_gaussian(entity: Entity, data_pos: Mat):
params = entity.training_params
# Positive phase # Positive phase
h_probs_pos = entity.h_given_v(data_pos) if params.do_batch_sample:
h_probs_pos = entity.h_given_v(sample_gaussian(data_pos))
else:
h_probs_pos = entity.h_given_v(data_pos)
# Sample hidden states # Sample hidden states
h_states_pos = h_probs_pos + sample_gaussian(h_probs_pos) h_probs_pos += sample_gaussian(h_probs_pos)
# Update weights (positive phase) # Update weights (positive phase)
dw = np.dot(np.transpose(data_pos), h_states_pos) dw = np.dot(np.transpose(data_pos), h_probs_pos)
dbh = np.sum(h_states_pos, 0) dbh = np.sum(h_probs_pos, 0)
dbv = np.sum(data_pos, 0) dbv = np.sum(data_pos, 0)
# Negative phase # Negative phase
data_neg = entity.v_given_h(h_states_pos) data_neg = entity.v_given_h(h_probs_pos)
h_probs_neg = entity.h_given_v(data_neg) h_probs_neg = entity.h_given_v(data_neg)
# Update weights (negative phase) # Update weights (negative phase)