tsne_manifold_compression

Bases: manifold_compression

A manifold-based dimensionality reduction class using t-SNE (t-distributed Stochastic Neighbor Embedding).

This class applies the t-SNE algorithm to reduce the dimensionality of input data, maintaining the local structure of the data in the reduced space.

Methods:

Name	Description
__init__	Initializes the t-SNE-based dimensionality reduction instance.

Parameters:

Name	Type	Description	Default
D	int	Number of dimensions to retain after compression.	required
perplexity	float	Perplexity parameter for the t-SNE algorithm, which controls the balance between local and global aspects of the data.	required
name	str	Name of the transformation. Defaults to 'tsne_manifold_compression'.	'tsne_manifold_compression'
*args	tuple	Additional positional arguments for the parent manifold_compression class.	()
**kwargs	dict	Additional keyword arguments for the parent manifold_compression class.	{}

Source code in tinybig/compression/manifold_compression.py

class tsne_manifold_compression(manifold_compression):
    """
        A manifold-based dimensionality reduction class using t-SNE (t-distributed Stochastic Neighbor Embedding).

        This class applies the t-SNE algorithm to reduce the dimensionality of input data, maintaining
        the local structure of the data in the reduced space.

        Methods
        -------
        __init__(D, perplexity, name='tsne_manifold_compression', *args, **kwargs)
            Initializes the t-SNE-based dimensionality reduction instance.

        Parameters
        ----------
        D : int
            Number of dimensions to retain after compression.
        perplexity : float
            Perplexity parameter for the t-SNE algorithm, which controls the balance between local and global
            aspects of the data.
        name : str, optional
            Name of the transformation. Defaults to 'tsne_manifold_compression'.
        *args : tuple
            Additional positional arguments for the parent `manifold_compression` class.
        **kwargs : dict
            Additional keyword arguments for the parent `manifold_compression` class.
    """
    def __init__(self, D: int, perplexity: float, name='tsne_manifold_compression', *args, **kwargs):
        """
            Initializes the t-SNE-based dimensionality reduction instance.

            Parameters
            ----------
            D : int
                Number of dimensions to retain after compression.
            perplexity : float
                Perplexity parameter for the t-SNE algorithm, which controls the balance between local and global
                aspects of the data.
            name : str, optional
                Name of the transformation. Defaults to 'tsne_manifold_compression'.
            *args : tuple
                Additional positional arguments for the parent `manifold_compression` class.
            **kwargs : dict
                Additional keyword arguments for the parent `manifold_compression` class.
        """
        manifold_function = tsne_manifold(n_components=D, perplexity=perplexity)
        super().__init__(D=D, name=name, manifold_function=manifold_function, *args, **kwargs)

__init__(D, perplexity, name='tsne_manifold_compression', *args, **kwargs)

Initializes the t-SNE-based dimensionality reduction instance.

Parameters:

Name	Type	Description	Default
D	int	Number of dimensions to retain after compression.	required
perplexity	float	Perplexity parameter for the t-SNE algorithm, which controls the balance between local and global aspects of the data.	required
name	str	Name of the transformation. Defaults to 'tsne_manifold_compression'.	'tsne_manifold_compression'
*args	tuple	Additional positional arguments for the parent manifold_compression class.	()
**kwargs	dict	Additional keyword arguments for the parent manifold_compression class.	{}

Source code in tinybig/compression/manifold_compression.py

def __init__(self, D: int, perplexity: float, name='tsne_manifold_compression', *args, **kwargs):
    """
        Initializes the t-SNE-based dimensionality reduction instance.

        Parameters
        ----------
        D : int
            Number of dimensions to retain after compression.
        perplexity : float
            Perplexity parameter for the t-SNE algorithm, which controls the balance between local and global
            aspects of the data.
        name : str, optional
            Name of the transformation. Defaults to 'tsne_manifold_compression'.
        *args : tuple
            Additional positional arguments for the parent `manifold_compression` class.
        **kwargs : dict
            Additional keyword arguments for the parent `manifold_compression` class.
    """
    manifold_function = tsne_manifold(n_components=D, perplexity=perplexity)
    super().__init__(D=D, name=name, manifold_function=manifold_function, *args, **kwargs)