main.py

import os
import argparse
import numpy as np
from torch.backends import cudnn
from utils.utils import *
from solver import Solver
import time
import warnings
import random

warnings.filterwarnings("ignore")

import sys
def setup_seed(seed):
     torch.manual_seed(seed)
     torch.cuda.manual_seed_all(seed)
     np.random.seed(seed)
     random.seed(seed)

class Logger(object):
    def __init__(self, filename="default.log", add_flag=True, stream=sys.stdout):
        self.terminal = stream
        self.filename = filename
        self.add_flag = add_flag

    def write(self, message):
        if self.add_flag:
            with open(self.filename, "a+") as log:
                self.terminal.write(message)
                log.write(message)
        else:
            with open(self.filename, "w") as log:
                self.terminal.write(message)
                log.write(message)

    def flush(self):
        pass


def str2bool(v):
    return v.lower() in ("true")


def find_nearest(array, value):
    array = np.asarray(array)
    idx = (np.abs(array - value)).argmin()
    return int(array[idx - 1])


def main(config):
    cudnn.benchmark = True
    if not os.path.exists(config.model_save_path):
        mkdir(config.model_save_path)
        if not os.path.exists(config.model_save_path+f"/{config.dataset}"):
            mkdir(config.model_save_path+f"/{config.dataset}")
            config.model_save_path = f"{config.model_save_path}/{config.dataset}"
    config.model_save_path = f"{config.model_save_path}/{config.dataset}"
    solver = Solver(vars(config))

    if config.mode == "train":
        solver.train()
    elif config.mode == "test":
        solver.test()

    return solver


if __name__ == "__main__":
    parser = argparse.ArgumentParser()

    # Alternative
    parser.add_argument("--win_size", type=int, default=100)
    parser.add_argument("--patch_size", type=list, default=[5])
    parser.add_argument("--lr", type=float, default=1e-4)
    parser.add_argument("--loss_fuc", type=str, default="MSE")
    parser.add_argument("--n_heads", type=int, default=1)
    parser.add_argument("--e_layers", type=int, default=3)
    parser.add_argument("--d_model", type=int, default=256)
    parser.add_argument("--rec_timeseries", action="store_true", default=True)

    parser.add_argument("--use_gpu", type=bool, default=True, help="use gpu")
    parser.add_argument("--gpu", type=int, default=0, help="gpu")
    parser.add_argument(
        "--use_multi_gpu", action="store_true", help="use multiple gpus", default=False
    )
    parser.add_argument(
        "--devices", type=str, default="1", help="device ids of multile gpus"
    )

    # Default
    parser.add_argument("--index", type=int, default=137)
    parser.add_argument("--num_epochs", type=int, default=10)
    parser.add_argument("--batch_size", type=int, default=128)
    parser.add_argument("--input_c", type=int, default=9)
    parser.add_argument("--output_c", type=int, default=9)
    parser.add_argument("--k", type=int, default=3)
    parser.add_argument("--dataset", type=str, default="credit")
    parser.add_argument("--mode", type=str, default="train", choices=["train", "test"])
    parser.add_argument("--data_path", type=str, default="./dataset/creditcard_ts.csv")
    parser.add_argument("--model_save_path", type=str, default="checkpoints")

    parser.add_argument("--anormly_ratio", type=float, default=4.00)

    config = parser.parse_args()
    args = vars(config)
    config.patch_size = [int(patch_index) for patch_index in config.patch_size]

    if config.dataset == "UCR":
        batch_size_buffer = [2, 4, 8, 16, 32, 64, 128, 256]
        data_len = np.load(
            "dataset/" + config.data_path + "/UCR_" + str(config.index) + "_train.npy"
        ).shape[0]
        config.batch_size = find_nearest(batch_size_buffer, data_len / config.win_size)
    elif config.dataset == "UCR_AUG":
        batch_size_buffer = [2, 4, 8, 16, 32, 64, 128, 256]
        data_len = np.load(
            "dataset/"
            + config.data_path
            + "/UCR_AUG_"
            + str(config.index)
            + "_train.npy"
        ).shape[0]
        config.batch_size = find_nearest(batch_size_buffer, data_len / config.win_size)
    elif config.dataset == "SMD_Ori":
        batch_size_buffer = [2, 4, 8, 16, 32, 64, 128, 256, 512]
        data_len = np.load(
            "dataset/"
            + config.data_path
            + "/SMD_Ori_"
            + str(config.index)
            + "_train.npy"
        ).shape[0]
        config.batch_size = find_nearest(batch_size_buffer, data_len / config.win_size)

    config.use_gpu = True if torch.cuda.is_available() and config.use_gpu else False
    # if config.use_gpu and config.use_multi_gpu:
    #     config.devices = config.devices.replace(" ", "")
        # device_ids = config.devices.split(",")
        # config.device_ids = [int(id_) for id_ in device_ids]
        # config.gpu = config.device_ids[0]
        # print(config.devices)
        # config.gpu = config.devices
        # os.environ['CUDA_VISIBLE_DEVICES']= config.devices #config.gpu

    sys.stdout = Logger("result/" + config.data_path + ".log", sys.stdout)
    if config.mode == "train":
        print("\n\n")
        print(time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()))
        print("================ Hyperparameters ===============")
        for k, v in sorted(args.items()):
            print("%s: %s" % (str(k), str(v)))
        print("====================  Train  ===================")
    setup_seed(1)
    main(config)