utils.py

import torch
import numpy as np
import random

def set_seed(seed):
    torch.manual_seed(seed)
    np.random.seed(seed)
    torch.cuda.manual_seed_all(seed)
    random.seed(seed)
    torch.backends.cudnn.deterministic = True
    torch.backends.cudnn.benchmark = False

def random_perturb(feature_len, length):
    r = np.linspace(0, feature_len, length + 1, dtype = np.uint16)
    return r

# def norm(data):
#     l2 = torch.norm(data, p = 2, dim = -1, keepdim = True)
#     return torch.div(data, l2)
    
def save_best_record(test_info, file_path):
    fo = open(file_path, "w")
    fo.write("Step: {}\n".format(test_info["step"][-1]))
    fo.write("auc: {:.4f}\n".format(test_info["auc"][-1]))
    fo.write("ap: {:.4f}\n".format(test_info["ap"][-1]))
    fo.write("loss: {:.4f}\n".format(test_info["loss"][-1]))
    fo.write("accuracy: {:.4f}\n".format(test_info["accuracy"][-1]))
    fo.close()


def process_feat(feat, length):
    new_feat = np.zeros((length, feat.shape[1])).astype(np.float32)

    r = np.linspace(0, len(feat), length+1, dtype=np.int32)
    for i in range(length):
        if r[i]!=r[i+1]:
            new_feat[i,:] = np.mean(feat[r[i]:r[i+1],:], 0)
        else:
            new_feat[i,:] = feat[r[i], :]
    return new_feat