SNN2.src.model.layers.CURELayer.algorithm module

TfCURE module

This module is used to compute the CURE algorithm on a dataset using tensorflow

class SNN2.src.model.layers.CURELayer.algorithm.TfCURE(points_set: Tensor, number_of_clusters: int, number_of_representatives: int = 5, compression_factor: float = 0.5, engine: str = 'Default')

Bases: object

TfCURE.

This class is the entry point to execute the CURE algorithm using tensorflow

property clusters: Tensor | None

create_queue(*, logger=None, write_msg=<function f_logger.<locals>.__dummy_log>, **kwargs) → None

process() → None

class SNN2.src.model.layers.CURELayer.algorithm.TfCluster(cluster: Tensor, c_id: int, dtype_int: Type = tf.int32, dtype_pt: Type = tf.float32)

Bases: object

TfCluster.

Class that manages a single Tensorflow cluster

compute_representors(n_representors: int = 1, compression_factor: float = 0.0) → None

get_furthest(adversaries: Tensor, reduction_function: Callable | None = None) → Tuple[Tensor, Tensor]

property mean: Tensor

min_reduction(t: RaggedTensor, axis: int = 0) → Tuple[Tensor, Tensor]

my_dst(adversaries: Tensor | Self, distance_function: Callable | None = None) → Tensor

class SNN2.src.model.layers.CURELayer.algorithm.TfClusters(clusters)

Bases: object