cogdl.data.sampler

Module Contents

Classes

Sampler

SAINTSampler

NodeSampler

EdgeSampler

RWSampler

MRWSampler

LayerSampler

class cogdl.data.sampler.Sampler(data, args_params)[source]
sample(self)[source]
class cogdl.data.sampler.SAINTSampler(data, args_params)[source]

Bases: cogdl.data.sampler.Sampler

estimate(self)[source]
gen_subgraph(self)[source]
sample(self)[source]
extract_subgraph(self, edge_idx, directed=True)[source]
get_subgraph(self, phase, require_norm=True)[source]

Generate one minibatch for model. In the ‘train’ mode, one minibatch corresponds to one subgraph of the training graph. In the ‘valid’ or ‘test’ mode, one batch corresponds to the full graph (i.e., full-batch rather than minibatch evaluation for validation / test sets).

Inputs:

mode str, can be ‘train’, ‘valid’, ‘test’ require_norm boolean

Outputs:

data Data object, modeling the sampled subgraph data.norm_aggr aggregation normalization data.norm_loss normalization normalization

class cogdl.data.sampler.NodeSampler(data, args_params)[source]

Bases: cogdl.data.sampler.SAINTSampler

sample(self)[source]
class cogdl.data.sampler.EdgeSampler(data, args_params)[source]

Bases: cogdl.data.sampler.SAINTSampler

sample(self)[source]
class cogdl.data.sampler.RWSampler(data, args_params)[source]

Bases: cogdl.data.sampler.SAINTSampler

sample(self)[source]
class cogdl.data.sampler.MRWSampler(data, args_params)[source]

Bases: cogdl.data.sampler.SAINTSampler

sample(self)[source]
class cogdl.data.sampler.LayerSampler(data, model, params_args)[source]

Bases: cogdl.data.sampler.Sampler

get_batches(self, train_nodes, train_labels, batch_size=64, shuffle=True)[source]