mask.BaseMask

Description

Base class for all masks. It serves as a boilerplate for other masks. It is not meant to be used directly.

Parameters

• dims (list, default: [1, 100, 1]):
  A list specifying the dimensions of the network in the order [input_dim, hidden_size, readout_dim].

• hidden_size (int):
  Number of neurons in the hidden layer, extracted from dims[1].

• input_dim (int, default: 1):
  Dimensionality of the input, extracted from dims[0].

• readout_dim (int, default: 1):
  Dimensionality of the output, extracted from dims[2].

Methods

get_input_indices()

Abstract method to return the indices of neurons that receive input.
Must be implemented in a child class.

Raises:
NotImplementedError if called directly.

get_non_input_indices()

Abstract method to return the indices of neurons that do not receive input.
Must be implemented in a child class.

Raises:
NotImplementedError if called directly.

get_readout_indices()

Abstract method to return the indices of neurons that send readout.
Must be implemented in a child class.

Raises:
NotImplementedError if called directly.

get_areas()

Abstract method to return the number of areas in the network.
Must be implemented in a child class.

Raises:
NotImplementedError if called directly.

get_area_indices(area)

Abstract method to return the indices of neurons in a specific area.
Must be implemented in a child class.

Parameters:

• area (int): The area identifier.

Raises:
NotImplementedError if called directly.

_generate_sparse_matrix(n, m, p)

Generates a sparse binary matrix of size (n, m) with density p.
A value of 1 indicates a connection, while 0 indicates no connection.

Parameters:

• n (int): Number of rows.
• m (int): Number of columns.
• p (float): Probability of a connection, must be in [0, 1].

Returns:

• mask (np.ndarray): Sparse binary matrix of shape (n, m).

Usage:

sparse_matrix = mask._generate_sparse_matrix(10, 20, 0.1)

plot_masks()

Plots the connectivity masks for:

• Input layer
• Hidden layer
• Readout layer

Usage:

mask.plot_masks()

get_masks()

Returns the transposed connectivity masks for:

• Input layer
• Hidden layer
• Readout layer

Assertions:

• The masks must already be generated.

Returns:

• List[np.ndarray]: A list of transposed masks for the input, hidden, and readout layers.

Usage:

input_mask, hidden_mask, readout_mask = mask.get_masks()

get_specs()

Returns a dictionary containing the specifications of the mask.

Returns:

• dict: Specifications including:
  • "dims"
  • "hidden_size"
  • "input_dim"
  • "readout_dim"

Usage:

specs = mask.get_specs()

print_specs()

Prints the specifications of the mask in a formatted manner.

Usage:

mask.print_specs()