buddi.models.buddi4

Functions

Input([shape, batch_size, dtype, sparse, ...])	Used to instantiate a Keras tensor.
build_buddi4(n_x, n_y, n_labels, n_stims, ...)	Builds the BUDDI model.
build_encoder_branch(inputs, hidden_dim, z_dim)	Defines a single hidden layer encoder that outputs two encodings: mu and log_var.
build_latent_space_classifier(inputs, ...[, ...])	A simple classifier that predicts from the latent space representation.
build_prop_estimator(inputs, num_classes[, ...])	Defines a classifier network that estimates cell type proportion from the input.
build_semi_supervised_decoder(...[, ...])	Defines a decoder network that takes in both supervised and unsupervised inputs.
classifier_loss_generator([weight, loss_fn, ...])	Classifier loss function generator.
fit_buddi4(supervised_model, ...[, epochs, ...])	Train buddi4 model with supervised and unsupervised datasets,
kl_loss_generator([beta, agg_fn])	KL divergence loss function generator.
reconstr_loss_generator([weight, ...])	Reconstruction loss function generator.
unsupervised_dummy_loss_fn(y_true, y_pred)	Dummy loss function for unsupervised branch proportion estimator.

Classes

Adam([learning_rate, beta_1, beta_2, ...])	Optimizer that implements the Adam algorithm.
CategoricalCrossentropy([from_logits, ...])	Computes the crossentropy loss between the labels and predictions.
Concatenate(*args, **kwargs)	Concatenates a list of inputs.
MeanSquaredError([reduction, name, dtype])	Computes the mean of squares of errors between labels and predictions.
Model(*args, **kwargs)	A model grouping layers into an object with training/inference features.
ReparameterizationLayer(*args, **kwargs)	Custom layer that applies the reparameterization trick.
tqdm(*_, **__)	Decorate an iterable object, returning an iterator which acts exactly like the original iterable, but prints a dynamically updating progressbar every time a value is requested.