pytorch_gats.py

# Copyright (c) Microsoft Corporation.
# Licensed under the MIT License.


from __future__ import division
from __future__ import print_function

import os
import numpy as np
import pandas as pd
import copy
from sklearn.metrics import roc_auc_score, mean_squared_error
import logging
from ...utils import (
    unpack_archive_with_buffer,
    save_multiple_parts_file,
    get_or_create_path,
    drop_nan_by_y_index,
)
from ...log import get_module_logger, TimeInspector
import torch
import torch.nn as nn
import torch.optim as optim

from .pytorch_utils import count_parameters
from ...model.base import Model
from ...data.dataset import DatasetH
from ...data.dataset.handler import DataHandlerLP
from ...contrib.model.pytorch_lstm import LSTMModel
from ...contrib.model.pytorch_gru import GRUModel


class GATs(Model):
    """GATs Model

    Parameters
    ----------
    lr : float
        learning rate
    d_feat : int
        input dimensions for each time step
    metric : str
        the evaluate metric used in early stop
    optimizer : str
        optimizer name
    GPU : int
        the GPU ID used for training
    """

    def __init__(
        self,
        d_feat=6,
        hidden_size=64,
        num_layers=2,
        dropout=0.0,
        n_epochs=200,
        lr=0.001,
        metric="",
        early_stop=20,
        loss="mse",
        base_model="GRU",
        with_pretrain=True,
        model_path=None,
        optimizer="adam",
        GPU=0,
        seed=None,
        **kwargs
    ):
        # Set logger.
        self.logger = get_module_logger("GATs")
        self.logger.info("GATs pytorch version...")

        # set hyper-parameters.
        self.d_feat = d_feat
        self.hidden_size = hidden_size
        self.num_layers = num_layers
        self.dropout = dropout
        self.n_epochs = n_epochs
        self.lr = lr
        self.metric = metric
        self.early_stop = early_stop
        self.optimizer = optimizer.lower()
        self.loss = loss
        self.base_model = base_model
        self.with_pretrain = with_pretrain
        self.model_path = model_path
        self.device = torch.device("cuda:%d" % (GPU) if torch.cuda.is_available() and GPU >= 0 else "cpu")
        self.use_gpu = torch.cuda.is_available()
        self.seed = seed

        self.logger.info(
            "GATs parameters setting:"
            "\nd_feat : {}"
            "\nhidden_size : {}"
            "\nnum_layers : {}"
            "\ndropout : {}"
            "\nn_epochs : {}"
            "\nlr : {}"
            "\nmetric : {}"
            "\nearly_stop : {}"
            "\noptimizer : {}"
            "\nloss_type : {}"
            "\nbase_model : {}"
            "\nwith_pretrain : {}"
            "\nmodel_path : {}"
            "\ndevice : {}"
            "\nuse_GPU : {}"
            "\nseed : {}".format(
                d_feat,
                hidden_size,
                num_layers,
                dropout,
                n_epochs,
                lr,
                metric,
                early_stop,
                optimizer.lower(),
                loss,
                base_model,
                with_pretrain,
                model_path,
                self.device,
                self.use_gpu,
                seed,
            )
        )

        if self.seed is not None:
            np.random.seed(self.seed)
            torch.manual_seed(self.seed)

        self.GAT_model = GATModel(
            d_feat=self.d_feat,
            hidden_size=self.hidden_size,
            num_layers=self.num_layers,
            dropout=self.dropout,
            base_model=self.base_model,
        )
        self.logger.info("model:\n{:}".format(self.GAT_model))
        self.logger.info("model size: {:.4f} MB".format(count_parameters(self.GAT_model)))

        if optimizer.lower() == "adam":
            self.train_optimizer = optim.Adam(self.GAT_model.parameters(), lr=self.lr)
        elif optimizer.lower() == "gd":
            self.train_optimizer = optim.SGD(self.GAT_model.parameters(), lr=self.lr)
        else:
            raise NotImplementedError("optimizer {} is not supported!".format(optimizer))

        self.fitted = False
        self.GAT_model.to(self.device)

    @property
    def use_gpu(self):
        return self.device != torch.device("cpu")

    def mse(self, pred, label):
        loss = (pred - label) ** 2
        return torch.mean(loss)

    def loss_fn(self, pred, label):
        mask = ~torch.isnan(label)

        if self.loss == "mse":
            return self.mse(pred[mask], label[mask])

        raise ValueError("unknown loss `%s`" % self.loss)

    def metric_fn(self, pred, label):

        mask = torch.isfinite(label)

        if self.metric == "" or self.metric == "loss":
            return -self.loss_fn(pred[mask], label[mask])

        raise ValueError("unknown metric `%s`" % self.metric)

    def get_daily_inter(self, df, shuffle=False):
        # organize the train data into daily batches
        daily_count = df.groupby(level=0).size().values
        daily_index = np.roll(np.cumsum(daily_count), 1)
        daily_index[0] = 0
        if shuffle:
            # shuffle data
            daily_shuffle = list(zip(daily_index, daily_count))
            np.random.shuffle(daily_shuffle)
            daily_index, daily_count = zip(*daily_shuffle)
        return daily_index, daily_count

    def train_epoch(self, x_train, y_train):

        x_train_values = x_train.values
        y_train_values = np.squeeze(y_train.values)
        self.GAT_model.train()

        # organize the train data into daily batches
        daily_index, daily_count = self.get_daily_inter(x_train, shuffle=True)

        for idx, count in zip(daily_index, daily_count):
            batch = slice(idx, idx + count)
            feature = torch.from_numpy(x_train_values[batch]).float().to(self.device)
            label = torch.from_numpy(y_train_values[batch]).float().to(self.device)

            pred = self.GAT_model(feature)
            loss = self.loss_fn(pred, label)

            self.train_optimizer.zero_grad()
            loss.backward()
            torch.nn.utils.clip_grad_value_(self.GAT_model.parameters(), 3.0)
            self.train_optimizer.step()

    def test_epoch(self, data_x, data_y):

        # prepare training data
        x_values = data_x.values
        y_values = np.squeeze(data_y.values)

        self.GAT_model.eval()

        scores = []
        losses = []

        # organize the test data into daily batches
        daily_index, daily_count = self.get_daily_inter(data_x, shuffle=False)

        for idx, count in zip(daily_index, daily_count):
            batch = slice(idx, idx + count)
            feature = torch.from_numpy(x_values[batch]).float().to(self.device)
            label = torch.from_numpy(y_values[batch]).float().to(self.device)

            pred = self.GAT_model(feature)
            loss = self.loss_fn(pred, label)
            losses.append(loss.item())

            score = self.metric_fn(pred, label)
            scores.append(score.item())

        return np.mean(losses), np.mean(scores)

    def fit(
        self,
        dataset: DatasetH,
        evals_result=dict(),
        save_path=None,
    ):

        df_train, df_valid, df_test = dataset.prepare(
            ["train", "valid", "test"],
            col_set=["feature", "label"],
            data_key=DataHandlerLP.DK_L,
        )

        x_train, y_train = df_train["feature"], df_train["label"]
        x_valid, y_valid = df_valid["feature"], df_valid["label"]

        save_path = get_or_create_path(save_path)
        stop_steps = 0
        best_score = -np.inf
        best_epoch = 0
        evals_result["train"] = []
        evals_result["valid"] = []

        # load pretrained base_model
        if self.with_pretrain:
            if self.model_path == None:
                raise ValueError("the path of the pretrained model should be given first!")
            self.logger.info("Loading pretrained model...")
            if self.base_model == "LSTM":
                pretrained_model = LSTMModel()
                pretrained_model.load_state_dict(torch.load(self.model_path))
            elif self.base_model == "GRU":
                pretrained_model = GRUModel()
                pretrained_model.load_state_dict(torch.load(self.model_path))
            else:
                raise ValueError("unknown base model name `%s`" % self.base_model)

            model_dict = self.GAT_model.state_dict()
            pretrained_dict = {k: v for k, v in pretrained_model.state_dict().items() if k in model_dict}
            model_dict.update(pretrained_dict)
            self.GAT_model.load_state_dict(model_dict)
            self.logger.info("Loading pretrained model Done...")

        # train
        self.logger.info("training...")
        self.fitted = True

        for step in range(self.n_epochs):
            self.logger.info("Epoch%d:", step)
            self.logger.info("training...")
            self.train_epoch(x_train, y_train)
            self.logger.info("evaluating...")
            train_loss, train_score = self.test_epoch(x_train, y_train)
            val_loss, val_score = self.test_epoch(x_valid, y_valid)
            self.logger.info("train %.6f, valid %.6f" % (train_score, val_score))
            evals_result["train"].append(train_score)
            evals_result["valid"].append(val_score)

            if val_score > best_score:
                best_score = val_score
                stop_steps = 0
                best_epoch = step
                best_param = copy.deepcopy(self.GAT_model.state_dict())
            else:
                stop_steps += 1
                if stop_steps >= self.early_stop:
                    self.logger.info("early stop")
                    break

        self.logger.info("best score: %.6lf @ %d" % (best_score, best_epoch))
        self.GAT_model.load_state_dict(best_param)
        torch.save(best_param, save_path)

        if self.use_gpu:
            torch.cuda.empty_cache()

    def predict(self, dataset):
        if not self.fitted:
            raise ValueError("model is not fitted yet!")

        x_test = dataset.prepare("test", col_set="feature")
        index = x_test.index
        self.GAT_model.eval()
        x_values = x_test.values
        preds = []

        # organize the data into daily batches
        daily_index, daily_count = self.get_daily_inter(x_test, shuffle=False)

        for idx, count in zip(daily_index, daily_count):
            batch = slice(idx, idx + count)
            x_batch = torch.from_numpy(x_values[batch]).float().to(self.device)

            with torch.no_grad():
                pred = self.GAT_model(x_batch).detach().cpu().numpy()

            preds.append(pred)

        return pd.Series(np.concatenate(preds), index=index)


class GATModel(nn.Module):
    def __init__(self, d_feat=6, hidden_size=64, num_layers=2, dropout=0.0, base_model="GRU"):
        super().__init__()

        if base_model == "GRU":
            self.rnn = nn.GRU(
                input_size=d_feat,
                hidden_size=hidden_size,
                num_layers=num_layers,
                batch_first=True,
                dropout=dropout,
            )
        elif base_model == "LSTM":
            self.rnn = nn.LSTM(
                input_size=d_feat,
                hidden_size=hidden_size,
                num_layers=num_layers,
                batch_first=True,
                dropout=dropout,
            )
        else:
            raise ValueError("unknown base model name `%s`" % base_model)

        self.hidden_size = hidden_size
        self.d_feat = d_feat
        self.transformation = nn.Linear(self.hidden_size, self.hidden_size)
        self.a = nn.Parameter(torch.randn(self.hidden_size * 2, 1))
        self.a.requires_grad = True
        self.fc = nn.Linear(self.hidden_size, self.hidden_size)
        self.fc_out = nn.Linear(hidden_size, 1)
        self.leaky_relu = nn.LeakyReLU()
        self.softmax = nn.Softmax(dim=1)

    def cal_attention(self, x, y):
        x = self.transformation(x)
        y = self.transformation(y)

        sample_num = x.shape[0]
        dim = x.shape[1]
        e_x = x.expand(sample_num, sample_num, dim)
        e_y = torch.transpose(e_x, 0, 1)
        attention_in = torch.cat((e_x, e_y), 2).view(-1, dim * 2)
        self.a_t = torch.t(self.a)
        attention_out = self.a_t.mm(torch.t(attention_in)).view(sample_num, sample_num)
        attention_out = self.leaky_relu(attention_out)
        att_weight = self.softmax(attention_out)
        return att_weight

    def forward(self, x):
        # x: [N, F*T]
        x = x.reshape(len(x), self.d_feat, -1)  # [N, F, T]
        x = x.permute(0, 2, 1)  # [N, T, F]
        out, _ = self.rnn(x)
        hidden = out[:, -1, :]
        att_weight = self.cal_attention(hidden, hidden)
        hidden = att_weight.mm(hidden) + hidden
        hidden = self.fc(hidden)
        hidden = self.leaky_relu(hidden)
        return self.fc_out(hidden).squeeze()