# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Some frequently used basic blocks
"""
import paddle
import paddle.nn as nn
[docs]class Activation(nn.Layer):
"""
Activation
"""
def __init__(self, act_type, **params):
super(Activation, self).__init__()
if act_type == 'relu':
self.act = nn.ReLU()
elif act_type == 'leaky_relu':
self.act = nn.LeakyReLU(**params)
else:
raise ValueError(act_type)
[docs] def forward(self, x):
"""tbd"""
return self.act(x)
[docs]class MLP(nn.Layer):
"""
MLP
"""
def __init__(self, layer_num, in_size, hidden_size, out_size, act, dropout_rate):
super(MLP, self).__init__()
layers = []
for layer_id in range(layer_num):
if layer_id == 0:
layers.append(nn.Linear(in_size, hidden_size))
layers.append(nn.Dropout(dropout_rate))
layers.append(Activation(act))
elif layer_id < layer_num - 1:
layers.append(nn.Linear(hidden_size, hidden_size))
layers.append(nn.Dropout(dropout_rate))
layers.append(Activation(act))
else:
layers.append(nn.Linear(hidden_size, out_size))
self.mlp = nn.Sequential(*layers)
[docs] def forward(self, x):
"""
Args:
x(tensor): (-1, dim).
"""
return self.mlp(x)
class RBF(nn.Layer):
"""
Radial Basis Function
"""
def __init__(self, centers, gamma, dtype='float32'):
super(RBF, self).__init__()
self.centers = paddle.reshape(paddle.to_tensor(centers, dtype=dtype), [1, -1])
self.gamma = gamma
def forward(self, x):
"""
Args:
x(tensor): (-1, 1).
Returns:
y(tensor): (-1, n_centers)
"""
x = paddle.reshape(x, [-1, 1])
return paddle.exp(-self.gamma * paddle.square(x - self.centers))