PFLlib/system/flcore/trainmodel/models.py

# PFLlib: Personalized Federated Learning Algorithm Library
# Copyright (C) 2021  Jianqing Zhang

# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.

# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.

# You should have received a copy of the GNU General Public License along
# with this program; if not, write to the Free Software Foundation, Inc.,
# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.

import torch
import torch.nn.functional as F
from torch import nn

batch_size = 10


# split an original model into a base and a head
class BaseHeadSplit(nn.Module):
    def __init__(self, base, head):
        super(BaseHeadSplit, self).__init__()

        self.base = base
        self.head = head

    def forward(self, x):
        out = self.base(x)
        out = self.head(out)

        return out


###########################################################


# https://github.com/jindongwang/Deep-learning-activity-recognition/blob/master/pytorch/network.py
class HARCNN(nn.Module):
    def __init__(
        self,
        in_channels=9,
        dim_hidden=64 * 26,
        num_classes=6,
        conv_kernel_size=(1, 9),
        pool_kernel_size=(1, 2),
    ):
        super().__init__()
        self.conv1 = nn.Sequential(
            nn.Conv2d(in_channels, 32, kernel_size=conv_kernel_size),
            nn.ReLU(),
            nn.MaxPool2d(kernel_size=pool_kernel_size, stride=2),
        )
        self.conv2 = nn.Sequential(
            nn.Conv2d(32, 64, kernel_size=conv_kernel_size),
            nn.ReLU(),
            nn.MaxPool2d(kernel_size=pool_kernel_size, stride=2),
        )
        self.fc = nn.Sequential(
            nn.Linear(dim_hidden, 1024),
            nn.ReLU(),
            nn.Linear(1024, 512),
            nn.ReLU(),
            nn.Linear(512, num_classes),
        )

    def forward(self, x):
        out = self.conv1(x)
        out = self.conv2(out)
        out = torch.flatten(out, 1)
        out = self.fc(out)
        return out


# https://github.com/FengHZ/KD3A/blob/master/model/digit5.py
class Digit5CNN(nn.Module):
    def __init__(self):
        super(Digit5CNN, self).__init__()
        self.encoder = nn.Sequential()
        self.encoder.add_module(
            "conv1", nn.Conv2d(3, 64, kernel_size=5, stride=1, padding=2)
        )
        self.encoder.add_module("bn1", nn.BatchNorm2d(64))
        self.encoder.add_module("relu1", nn.ReLU())
        self.encoder.add_module(
            "maxpool1",
            nn.MaxPool2d(kernel_size=3, stride=2, padding=1, ceil_mode=False),
        )
        self.encoder.add_module(
            "conv2", nn.Conv2d(64, 64, kernel_size=5, stride=1, padding=2)
        )
        self.encoder.add_module("bn2", nn.BatchNorm2d(64))
        self.encoder.add_module("relu2", nn.ReLU())
        self.encoder.add_module(
            "maxpool2",
            nn.MaxPool2d(kernel_size=3, stride=2, padding=1, ceil_mode=False),
        )
        self.encoder.add_module(
            "conv3", nn.Conv2d(64, 128, kernel_size=5, stride=1, padding=2)
        )
        self.encoder.add_module("bn3", nn.BatchNorm2d(128))
        self.encoder.add_module("relu3", nn.ReLU())

        self.linear = nn.Sequential()
        self.linear.add_module("fc1", nn.Linear(8192, 3072))
        self.linear.add_module("bn4", nn.BatchNorm1d(3072))
        self.linear.add_module("relu4", nn.ReLU())
        self.linear.add_module("dropout", nn.Dropout())
        self.linear.add_module("fc2", nn.Linear(3072, 2048))
        self.linear.add_module("bn5", nn.BatchNorm1d(2048))
        self.linear.add_module("relu5", nn.ReLU())

        self.fc = nn.Linear(2048, 10)

    def forward(self, x):
        batch_size = x.size(0)
        feature = self.encoder(x)
        feature = feature.view(batch_size, -1)
        feature = self.linear(feature)
        out = self.fc(feature)
        return out


# https://github.com/FengHZ/KD3A/blob/master/model/amazon.py
class AmazonMLP(nn.Module):
    def __init__(self):
        super(AmazonMLP, self).__init__()
        self.encoder = nn.Sequential(
            nn.Linear(5000, 1000),
            # nn.BatchNorm1d(1000),
            nn.ReLU(),
            nn.Linear(1000, 500),
            # nn.BatchNorm1d(500),
            nn.ReLU(),
            nn.Linear(500, 100),
            # nn.BatchNorm1d(100),
            nn.ReLU(),
        )
        self.fc = nn.Linear(100, 2)

    def forward(self, x):
        out = self.encoder(x)
        out = self.fc(out)
        return out


# # https://github.com/katsura-jp/fedavg.pytorch/blob/master/src/models/cnn.py
# class FedAvgCNN(nn.Module):
#     def __init__(self, in_features=1, num_classes=10, dim=1024):
#         super().__init__()
#         self.conv1 = nn.Conv2d(in_features,
#                                32,
#                                kernel_size=5,
#                                padding=0,
#                                stride=1,
#                                bias=True)
#         self.conv2 = nn.Conv2d(32,
#                                64,
#                                kernel_size=5,
#                                padding=0,
#                                stride=1,
#                                bias=True)
#         self.fc1 = nn.Linear(dim, 512)
#         self.fc = nn.Linear(512, num_classes)

#         self.act = nn.ReLU(inplace=True)
#         self.maxpool = nn.MaxPool2d(kernel_size=(2, 2))

#     def forward(self, x):
#         x = self.act(self.conv1(x))
#         x = self.maxpool(x)
#         x = self.act(self.conv2(x))
#         x = self.maxpool(x)
#         x = torch.flatten(x, 1)
#         x = self.act(self.fc1(x))
#         x = self.fc(x)
#         return x


class FedAvgCNN(nn.Module):
    def __init__(self, in_features=1, num_classes=10, dim=1024):
        super().__init__()
        self.conv1 = nn.Sequential(
            nn.Conv2d(
                in_features, 32, kernel_size=5, padding=0, stride=1, bias=True
            ),
            nn.ReLU(inplace=True),
            nn.MaxPool2d(kernel_size=(2, 2)),
        )
        self.conv2 = nn.Sequential(
            nn.Conv2d(32, 64, kernel_size=5, padding=0, stride=1, bias=True),
            nn.ReLU(inplace=True),
            nn.MaxPool2d(kernel_size=(2, 2)),
        )
        self.fc1 = nn.Sequential(nn.Linear(dim, 512), nn.ReLU(inplace=True))
        self.fc = nn.Linear(512, num_classes)

    def forward(self, x):
        out = self.conv1(x)
        out = self.conv2(out)
        out = torch.flatten(out, 1)
        out = self.fc1(out)
        out = self.fc(out)
        return out


# ====================================================================================================================


# https://github.com/katsura-jp/fedavg.pytorch/blob/master/src/models/mlp.py
class FedAvgMLP(nn.Module):
    def __init__(self, in_features=784, num_classes=10, hidden_dim=200):
        super().__init__()
        self.fc1 = nn.Linear(in_features, hidden_dim)
        self.fc2 = nn.Linear(hidden_dim, num_classes)
        self.act = nn.ReLU(inplace=True)

    def forward(self, x):
        if x.ndim == 4:
            x = x.view(x.size(0), -1)
        x = self.act(self.fc1(x))
        x = self.fc2(x)
        return x


# ====================================================================================================================


class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = nn.Conv2d(1, batch_size, 2, 1)
        self.conv2 = nn.Conv2d(batch_size, 32, 2, 1)
        self.dropout1 = nn.Dropout(0.25)
        self.dropout2 = nn.Dropout(0.5)
        self.fc1 = nn.Linear(18432, 128)
        self.fc = nn.Linear(128, 10)

    def forward(self, x):
        x = self.conv1(x)
        x = nn.ReLU()(x)
        x = nn.MaxPool2d(2, 1)(x)
        x = self.dropout1(x)
        x = self.conv2(x)
        x = nn.ReLU()(x)
        x = nn.MaxPool2d(2, 1)(x)
        x = self.dropout2(x)
        x = torch.flatten(x, 1)
        x = self.fc1(x)
        x = nn.ReLU()(x)
        x = self.fc(x)
        output = F.log_softmax(x, dim=1)
        return output


# ====================================================================================================================


class Mclr_Logistic(nn.Module):
    def __init__(self, input_dim=1 * 28 * 28, num_classes=10):
        super(Mclr_Logistic, self).__init__()
        self.fc = nn.Linear(input_dim, num_classes)

    def forward(self, x):
        x = torch.flatten(x, 1)
        x = self.fc(x)
        output = F.log_softmax(x, dim=1)
        return output


# ====================================================================================================================


class DNN(nn.Module):
    def __init__(self, input_dim=1 * 28 * 28, mid_dim=100, num_classes=10):
        super(DNN, self).__init__()
        self.fc1 = nn.Linear(input_dim, mid_dim)
        self.fc = nn.Linear(mid_dim, num_classes)

    def forward(self, x):
        x = torch.flatten(x, 1)
        x = F.relu(self.fc1(x))
        x = self.fc(x)
        x = F.log_softmax(x, dim=1)
        return x


# ====================================================================================================================


class CifarNet(nn.Module):
    def __init__(self, num_classes=10):
        super(CifarNet, self).__init__()
        self.conv1 = nn.Conv2d(3, 6, 5)
        self.pool = nn.MaxPool2d(2, 2)
        self.conv2 = nn.Conv2d(6, batch_size, 5)
        self.fc1 = nn.Linear(batch_size * 5 * 5, 120)
        self.fc2 = nn.Linear(120, 84)
        self.fc = nn.Linear(84, num_classes)

    def forward(self, x):
        x = self.pool(F.relu(self.conv1(x)))
        x = self.pool(F.relu(self.conv2(x)))
        x = x.view(-1, batch_size * 5 * 5)
        x = F.relu(self.fc1(x))
        x = F.relu(self.fc2(x))
        x = self.fc(x)
        x = F.log_softmax(x, dim=1)
        return x


# ====================================================================================================================

# cfg = {
#     'VGG11': [64, 'M', 128, 'M', 256, 256, 'M', 512, 512, 'M', 512, 512, 'M'],
#     'VGG13': [64, 64, 'M', 128, 128, 'M', 256, 256, 'M', 512, 512, 'M', 512, 512, 'M'],
#     'VGGbatch_size': [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 'M', 512, 512, 512, 'M', 512, 512, 512, 'M'],
#     'VGG19': [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 256, 'M', 512, 512, 512, 512, 'M', 512, 512, 512, 512, 'M'],
# }

# class VGG(nn.Module):
#     def __init__(self, vgg_name):
#         super(VGG, self).__init__()
#         self.features = self._make_layers(cfg[vgg_name])
#         self.classifier = nn.Sequential(
#             nn.Linear(512, 512),
#             nn.ReLU(True),
#             nn.Linear(512, 512),
#             nn.ReLU(True),
#             nn.Linear(512, 10)
#         )

#     def forward(self, x):
#         out = self.features(x)
#         out = out.view(out.size(0), -1)
#         out = self.classifier(out)
#         output = F.log_softmax(out, dim=1)
#         return output

#     def _make_layers(self, cfg):
#         layers = []
#         in_channels = 3
#         for x in cfg:
#             if x == 'M':
#                 layers += [nn.MaxPool2d(kernel_size=2, stride=2)]
#             else:
#                 layers += [nn.Conv2d(in_channels, x, kernel_size=3, padding=1),
#                            nn.BatchNorm2d(x),
#                            nn.ReLU(inplace=True)]
#                 in_channels = x
#         layers += [nn.AvgPool2d(kernel_size=1, stride=1)]
#         return nn.Sequential(*layers)

# ====================================================================================================================


def init_weights(m):
    classname = m.__class__.__name__
    if (
        classname.find("Conv2d") != -1
        or classname.find("ConvTranspose2d") != -1
    ):
        nn.init.kaiming_uniform_(m.weight)
        nn.init.zeros_(m.bias)
    elif classname.find("BatchNorm") != -1:
        nn.init.normal_(m.weight, 1.0, 0.02)
        nn.init.zeros_(m.bias)
    elif classname.find("Linear") != -1:
        nn.init.xavier_normal_(m.weight)
        nn.init.zeros_(m.bias)


class LeNet(nn.Module):
    def __init__(
        self,
        feature_dim=50 * 4 * 4,
        bottleneck_dim=256,
        num_classes=10,
        iswn=None,
    ):
        super(LeNet, self).__init__()

        self.conv_params = nn.Sequential(
            nn.Conv2d(1, 20, kernel_size=5),
            nn.MaxPool2d(2),
            nn.ReLU(),
            nn.Conv2d(20, 50, kernel_size=5),
            nn.Dropout2d(p=0.5),
            nn.MaxPool2d(2),
            nn.ReLU(),
        )
        self.bn = nn.BatchNorm1d(bottleneck_dim, affine=True)
        self.dropout = nn.Dropout(p=0.5)
        self.bottleneck = nn.Linear(feature_dim, bottleneck_dim)
        self.bottleneck.apply(init_weights)
        self.fc = nn.Linear(bottleneck_dim, num_classes)
        if iswn == "wn":
            self.fc = nn.utils.weight_norm(self.fc, name="weight")
        self.fc.apply(init_weights)

    def forward(self, x):
        x = self.conv_params(x)
        x = x.view(x.size(0), -1)
        x = self.bottleneck(x)
        x = self.bn(x)
        x = self.dropout(x)
        x = self.fc(x)
        x = F.log_softmax(x, dim=1)
        return x


# ====================================================================================================================

# class CNNCifar(nn.Module):
#     def __init__(self, num_classes=10):
#         super(CNNCifar, self).__init__()
#         self.conv1 = nn.Conv2d(3, 6, 5)
#         self.pool = nn.MaxPool2d(2, 2)
#         self.conv2 = nn.Conv2d(6, batch_size, 5)
#         self.fc1 = nn.Linear(batch_size * 5 * 5, 120)
#         self.fc2 = nn.Linear(120, 100)
#         self.fc3 = nn.Linear(100, num_classes)

#         # self.weight_keys = [['fc1.weight', 'fc1.bias'],
#         #                     ['fc2.weight', 'fc2.bias'],
#         #                     ['fc3.weight', 'fc3.bias'],
#         #                     ['conv2.weight', 'conv2.bias'],
#         #                     ['conv1.weight', 'conv1.bias'],
#         #                     ]

#     def forward(self, x):
#         x = self.pool(F.relu(self.conv1(x)))
#         x = self.pool(F.relu(self.conv2(x)))
#         x = x.view(-1, batch_size * 5 * 5)
#         x = F.relu(self.fc1(x))
#         x = F.relu(self.fc2(x))
#         x = self.fc3(x)
#         x = F.log_softmax(x, dim=1)
#         return x

# ====================================================================================================================


class LSTMNet(nn.Module):
    def __init__(
        self,
        hidden_dim,
        num_layers=2,
        bidirectional=False,
        dropout=0.2,
        padding_idx=0,
        vocab_size=98635,
        num_classes=10,
    ):
        super().__init__()

        self.dropout = nn.Dropout(dropout)
        self.embedding = nn.Embedding(vocab_size, hidden_dim, padding_idx)
        self.lstm = nn.LSTM(
            input_size=hidden_dim,
            hidden_size=hidden_dim,
            num_layers=num_layers,
            bidirectional=bidirectional,
            dropout=dropout,
            batch_first=True,
        )
        dims = hidden_dim * 2 if bidirectional else hidden_dim
        self.fc = nn.Linear(dims, num_classes)

    def forward(self, x):
        text, text_lengths = x

        embedded = self.embedding(text)

        # pack sequence
        packed_embedded = nn.utils.rnn.pack_padded_sequence(
            embedded, text_lengths, batch_first=True, enforce_sorted=False
        )
        packed_output, (hidden, cell) = self.lstm(packed_embedded)

        # unpack sequence
        out, out_lengths = nn.utils.rnn.pad_packed_sequence(
            packed_output, batch_first=True
        )

        out = torch.relu_(out[:, -1, :])
        out = self.dropout(out)
        out = self.fc(out)
        out = F.log_softmax(out, dim=1)

        return out


# ====================================================================================================================


class fastText(nn.Module):
    def __init__(
        self, hidden_dim, padding_idx=0, vocab_size=98635, num_classes=10
    ):
        super(fastText, self).__init__()

        # Embedding Layer
        self.embedding = nn.Embedding(vocab_size, hidden_dim, padding_idx)

        # Hidden Layer
        self.fc1 = nn.Linear(hidden_dim, hidden_dim)

        # Output Layer
        self.fc = nn.Linear(hidden_dim, num_classes)

    def forward(self, x):
        text, text_lengths = x

        embedded_sent = self.embedding(text)
        h = self.fc1(embedded_sent.mean(1))
        z = self.fc(h)
        out = F.log_softmax(z, dim=1)

        return out


# ====================================================================================================================


class TextCNN(nn.Module):
    def __init__(
        self,
        hidden_dim,
        num_channels=100,
        kernel_size=[3, 4, 5],
        max_len=200,
        dropout=0.8,
        padding_idx=0,
        vocab_size=98635,
        num_classes=10,
    ):
        super(TextCNN, self).__init__()

        # Embedding Layer
        self.embedding = nn.Embedding(vocab_size, hidden_dim, padding_idx)

        # This stackoverflow thread clarifies how conv1d works
        # https://stackoverflow.com/questions/46503816/keras-conv1d-layer-parameters-filters-and-kernel-size/46504997
        self.conv1 = nn.Sequential(
            nn.Conv1d(
                in_channels=hidden_dim,
                out_channels=num_channels,
                kernel_size=kernel_size[0],
            ),
            nn.ReLU(),
            nn.MaxPool1d(max_len - kernel_size[0] + 1),
        )
        self.conv2 = nn.Sequential(
            nn.Conv1d(
                in_channels=hidden_dim,
                out_channels=num_channels,
                kernel_size=kernel_size[1],
            ),
            nn.ReLU(),
            nn.MaxPool1d(max_len - kernel_size[1] + 1),
        )
        self.conv3 = nn.Sequential(
            nn.Conv1d(
                in_channels=hidden_dim,
                out_channels=num_channels,
                kernel_size=kernel_size[2],
            ),
            nn.ReLU(),
            nn.MaxPool1d(max_len - kernel_size[2] + 1),
        )

        self.dropout = nn.Dropout(dropout)

        # Fully-Connected Layer
        self.fc = nn.Linear(num_channels * len(kernel_size), num_classes)

    def forward(self, x):
        text, text_lengths = x

        embedded_sent = self.embedding(text).permute(0, 2, 1)

        conv_out1 = self.conv1(embedded_sent).squeeze(2)
        conv_out2 = self.conv2(embedded_sent).squeeze(2)
        conv_out3 = self.conv3(embedded_sent).squeeze(2)

        all_out = torch.cat((conv_out1, conv_out2, conv_out3), 1)
        final_feature_map = self.dropout(all_out)
        out = self.fc(final_feature_map)
        out = F.log_softmax(out, dim=1)

        return out


# ====================================================================================================================


# class linear(Function):
#   @staticmethod
#   def forward(ctx, input):
#     return input

#   @staticmethod
#   def backward(ctx, grad_output):
#     return grad_output