def __init__(self,
n_input_feat,
n_output_feat,
n_size,
init='glorot_uniform',
activation='relu',
activation_first=True,
**kwargs):
"""
Parameters
----------
n_input_feat: int
Number of input channels
n_output_feat: int
Number of output channels
n_size: int
Number of filter size(full length)
init: str, optional
Weight initialization for filters.
activation: str, optional
Activation function applied
activation_first: bool, optional
If to apply activation before convolution
"""
self.init = initializations.get(init) # Set weight initialization
self.activation = activations.get(activation) # Get activations
self.n_input_feat = n_input_feat
self.n_output_feat = n_output_feat
self.n_size = n_size
self.activation_first = activation_first
super(Conv2DUp, self).__init__(**kwargs)
def __init__(self,
n_input_feat,
n_output=2,
init='glorot_uniform',
activation='relu',
**kwargs):
"""
Parameters
----------
n_input_feat: int
Number of input channels
n_output: int, optional
Number of output channels: 2 for classification, 1 for regression
init: str, optional
Weight initialization for filters.
activation: str, optional
Activation function applied
"""
self.n_input_feat = n_input_feat
self.n_output = n_output
self.init = initializations.get(init) # Set weight initialization
self.activation = activations.get(activation) # Get activations
super(ContactMapGather, self).__init__(**kwargs)
def __init__(self,
n_input_feat,
n_output_feat,
n_size=3,
rate=[6, 12, 18, 24],
init='glorot_uniform',
activation='relu',
**kwargs):
"""
Parameters
----------
n_input_feat: int
Number of input channels
n_output_feat: int
Number of output channels for each Atrous component
n_size: int
Number of filter size(full length)
rate: int
Rate of each atrous convolution
init: str, optional
Weight initialization for filters.
activation: str, optional
Activation function applied
activation_first: bool, optional
If to apply activation before convolution
"""
self.init = initializations.get(init) # Set weight initialization
self.activation = activations.get(activation) # Get activations
self.n_input_feat = n_input_feat
self.n_output_feat = n_output_feat
self.n_size = n_size
self.rate = rate
super(Conv2DASPP, self).__init__(**kwargs)
def __init__(self,
pos_start=0,
pos_end=25,
embedding_length=50,
init='glorot_uniform',
activation='relu',
**kwargs):
"""
Parameters
----------
pos_start: int, optional
starting position of raw features that need embedding
pos_end: int, optional
ending position
embedding_length: int, optional
length for embedding
init: str, optional
Weight initialization for filters.
activation: str, optional
Activation function applied
"""
self.init = initializations.get(init) # Set weight initialization
self.activation = activations.get(activation) # Get activations
self.pos_start = pos_start
self.pos_end = pos_end
self.embedding_length = embedding_length
super(ResidueEmbedding, self).__init__(**kwargs)
def __init__(self,
n_input_feat,
n_output_feat,
n_size,
init='glorot_uniform',
activation='relu',
**kwargs):
"""
Parameters
----------
n_input_feat: int
Number of input channels
n_output_feat: int
Number of output channels
n_size: int
Number of filter size(full length)
init: str, optional
Weight initialization for filters.
activation: str, optional
Activation function applied
"""
self.init = initializations.get(init) # Set weight initialization
self.activation = activations.get(activation) # Get activations
self.n_input_feat = n_input_feat
self.n_output_feat = n_output_feat
self.n_size = n_size
super(Conv2DLayer_RaptorX, self).__init__(**kwargs)
def __init__(self,
n_input_feat,
n_output_feat,
n_size,
rate,
init='glorot_uniform',
activation='relu',
activation_first=True,
dropout=None,
**kwargs):
"""
Parameters
----------
n_input_feat: int
Number of input channels
n_output_feat: int
Number of output channels
n_size: int
Number of filter size(full length)
rate: int
Rate of atrous convolution
init: str, optional
Weight initialization for filters.
activation: str, optional
Activation function applied
dropout: float, optional
Dropout probability, not supported here
"""
self.init = initializations.get(init) # Set weight initialization
self.activation = activations.get(activation) # Get activations
self.n_input_feat = n_input_feat
self.n_output_feat = n_output_feat
self.n_size = n_size
self.rate = rate
self.activation_first = activation_first
super(DiagConv2DAtrous, self).__init__(**kwargs)