GURPP/train_gurp.py at main · luck-seu/GURPP · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
import random
import logging
import os
import sys

import numpy as np
import torch
from datetime import datetime

from tqdm import tqdm

from downstream_task import predict_crime, predict_check
from load_graph_data import HeteroGraphData, GURPData
from model.gurp import GURPModel
from gurpp_args import get_default_arguments

def get_logger(log_dir, name, log_filename='info.log', level=logging.INFO):
    logger = logging.getLogger(name)
    logger.setLevel(level)
    formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
    file_handler = logging.FileHandler(os.path.join(log_dir, log_filename))
    file_handler.setFormatter(formatter)
    console_formatter = logging.Formatter('%(asctime)s - %(levelname)s - %(message)s')
    console_handler = logging.StreamHandler(sys.stdout)
    console_handler.setFormatter(console_formatter)
    logger.addHandler(file_handler)
    logger.addHandler(console_handler)
    logger.info('Log directory: %s', log_dir)
    return logger

# Set up arguments
args = get_default_arguments()
training_key = 'gurp_training'

# Set up random seed
seed = args[training_key]['seed']
torch.manual_seed(seed=seed)
np.random.seed(seed)
random.seed(seed)
torch.cuda.manual_seed(seed)
torch.cuda.manual_seed_all(seed)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False

# Set up logger
log_folder = args[training_key]['log_folder']
logger_name = args[training_key]['logger_name']
log_level = args[training_key]['log_level']
log_attr = args[training_key]['log_attr']
log_file_name = logger_name + '_' + log_attr + '_' + datetime.now().strftime('%Y-%m-%d-%H-%M-%S')
logger = get_logger(log_folder, logger_name, log_file_name + '.log', level=log_level)

# Load data and init features
device = torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')
hkg = HeteroGraphData(kg_dir=args['data']['kg_dir'],
                      reverse=args['data']['kg_reverse'],
                      export=args['data']['kg_export'],
                      graph_device=device)
logger.info('kg data loaded...')
hkg.init_hetero_graph_features(image=args['data']['image'], flow=args['data']['flow'],
                               feature_dir=args['data']['hg_feature_dir'],
                               node_feats_dim=args['gurp_model']['node_dim'][0],
                               device=device)
logger.info('feature loaded...')

gurp_data = GURPData(hkg, if_test=False)
gurp_model = GURPModel(args['gurp_model']).to(device)
optimizer = torch.optim.Adam(gurp_model.parameters(), lr=args[training_key]['lr'], weight_decay=args[training_key]['weight_decay'])

def get_batch_mobility(mobility, batch_size, batch_num):
    mobility_batch = []
    for i in range(batch_num):
        cur_batch_mob = mobility[i*batch_size:(i+1)*batch_size]
        cur_batch_mob = cur_batch_mob[:, i*batch_size:(i+1)*batch_size]
        mobility_batch.append(cur_batch_mob)
    return mobility_batch

def train_model(args):
    epochs = args['epochs']
    batch_size = args['batch_size']
    batch_num = 180 // batch_size
    train_time = datetime.now().strftime('%Y-%m-%d_%H-%M-%S')

    best_mae_crime = 10000
    best_rmse_crime = 10000
    best_r2_crime = 0
    best_epoch_crime = 0

    sp_pos_all, sp_neg_all = gurp_data.get_all_samples(args['ratio_dict'], args['seed'])
    logger.info('samples loaded...')

    mobility = np.load(args['mobility'])
    mobility = torch.tensor(mobility, dtype=torch.float32).to(device)
    mob_batch = get_batch_mobility(mobility, batch_size, batch_num)
    reg_sub_b, sp_pos_b, sp_neg_b, reg_inf, reg_ouf, img = gurp_data.get_batch_samples(sp_pos_all, sp_neg_all,
                                                                                       batch_size, args)

    for epoch in tqdm(range(args['epochs'])):
        gurp_model.train()
        loss_epoch = []
        for batch in range(batch_num):
            optimizer.zero_grad()
            loss, sp_loss, flow_loss, img_loss, pred_loss, out_emb = gurp_model(reg_sub_b[batch], sp_pos_b[batch],
                                                                                sp_neg_b[batch], reg_inf[batch],
                                                                                reg_ouf[batch], mob_batch[batch],
                                                                                img[batch], batch_size, args)
            loss.backward()
            optimizer.step()
            loss_epoch.append(loss.item())
            logger.info(f"Epoch [{epoch}/{args['epochs']}]"
                        f"\t sp_loss: {sp_loss.item()} "
                        f"\t flow_loss: {flow_loss.item()} "
                        f"\t img_loss: {img_loss.item()}"
                        f"\t pred_loss: {pred_loss.item()}"
                        f"\t loss: {loss.item()}")

        with torch.no_grad():
            cur_out_emb = out_emb.detach().cpu().numpy()
            cri_mae, cri_rmse, cri_r2 = predict_crime(cur_out_emb)
            if cri_rmse < best_rmse_crime and cri_mae < best_mae_crime and best_r2_crime < cri_r2:
                best_rmse_crime = cri_rmse
                best_mae_crime = cri_mae
                best_r2_crime = cri_r2
                best_epoch_crime = epoch
                best_emb_crime = out_emb

    np.save(args['save_model_path'] + f'/region_emb_epoch_{epochs}_time_{train_time}.npy', cur_out_emb)
    logger.info('best epoch: {}, best crime result, cri_mae: {}, cri_rmse: {}, cri_r2: {}'.format(best_epoch_crime, best_mae_crime, best_rmse_crime, best_r2_crime))

    gurp_model.eval()
    with torch.no_grad():
        cri_mae, cri_rmse, cri_r2 = predict_crime(cur_out_emb)
        chk_mae, chk_rmse, chk_r2 = predict_check(cur_out_emb)
        logger.info('crime final result, cri_mae: {}, cri_rmse: {}, cri_r2: {}'.format(cri_mae, cri_rmse, cri_r2))
        logger.info('check final result, chk_mae: {}, chk_rmse: {}, chk_r2: {}'.format(chk_mae, chk_rmse, chk_r2))
        cri_mae, cri_rmse, cri_r2 = predict_crime(best_emb_crime.detach().cpu().numpy())
        chk_mae, chk_rmse, chk_r2 = predict_check(best_emb_crime.detach().cpu().numpy())
        logger.info('best final result, crime, mae: {}, rmse: {}, r2: {}'.format(cri_mae, cri_rmse, cri_r2))
        logger.info('best final result, check, mae: {}, rmse: {}, r2: {}'.format(chk_mae, chk_rmse, chk_r2))

train_model(args[training_key])