def test_synthetic_1():
synthetic_dataset = SyntheticDataset()
svaec = SVAEC(synthetic_dataset.nb_genes, synthetic_dataset.n_batches,
synthetic_dataset.n_labels)
infer_synthetic_svaec = JointSemiSupervisedVariationalInference(
svaec, synthetic_dataset, use_cuda=use_cuda)
infer_synthetic_svaec.train(n_epochs=1)
infer_synthetic_svaec.entropy_batch_mixing('labelled')
infer_synthetic_svaec.show_t_sne('labelled', n_samples=50)
infer_synthetic_svaec.show_t_sne('unlabelled',
n_samples=50,
color_by='labels')
infer_synthetic_svaec.show_t_sne('labelled',
n_samples=50,
color_by='batches and labels')
infer_synthetic_svaec.clustering_scores('labelled')
def test_synthetic_1():
synthetic_dataset = SyntheticDataset()
svaec = SVAEC(synthetic_dataset.nb_genes, synthetic_dataset.n_batches,
synthetic_dataset.n_labels)
infer_synthetic_svaec = JointSemiSupervisedVariationalInference(
svaec, synthetic_dataset, use_cuda=use_cuda)
infer_synthetic_svaec.fit(n_epochs=1)
infer_synthetic_svaec.entropy_batch_mixing('labelled')
vaec = VAEC(synthetic_dataset.nb_genes, synthetic_dataset.n_batches,
synthetic_dataset.n_labels)
infer_synthetic_vaec = JointSemiSupervisedVariationalInference(
vaec,
synthetic_dataset,
use_cuda=use_cuda,
early_stopping_metric='ll',
frequency=1,
save_best_state_metric='accuracy',
on='labelled')
infer_synthetic_vaec.fit(n_epochs=20)
infer_synthetic_vaec.svc_rf(unit_test=True)
infer_synthetic_vaec.show_t_sne('labelled', n_samples=50)