FLARE/dataLoader_my.py at main · CASIA-LMC-Lab/FLARE · 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
142
143
144
145
146
147
148
149
150
from torch.utils.data import DataLoader,Dataset
import numpy as np
import os
import pandas as pd
import torch
from typing import Callable
from sklearn.preprocessing import StandardScaler

class MyDataset1_noscaler(Dataset):
    def __init__(self, npy_dir_path,meta_feats_path,
                 kids_path, p2n_rate_train,p2n_rate_test,
                 data_split="train",if_only_metavec=False):
        super(MyDataset1_noscaler,self).__init__()
        self.npy_dir_path=npy_dir_path
        self.kids=np.load(kids_path)
        self.if_only_metavec=if_only_metavec
        if data_split=="train":
            self.ifTrain=True
        else:
            self.ifTrain=False
        self.load_meta_idxs_feats(meta_feats_path)
        self.p2n_rate_train=p2n_rate_train
        self.p2n_rate_test=p2n_rate_test
        self.data_split=data_split
        self._generate_train_()

    def load_meta_idxs_feats(self,meta_feats_path):
        data=np.load(meta_feats_path)
        meta_kids=data[:,0].flatten()
        meta_feats=data[:,1:].astype(np.float_)
        mask=np.isin(self.kids,meta_kids)
        if self.if_only_metavec:
            mean_=np.nanmean(meta_feats,axis=0)
            std_=np.nanstd(meta_feats,axis=0)
            meta_feats = (meta_feats-mean_)/std_
            meta_feats[np.isnan(meta_feats)]=0
        mask=np.isin(meta_kids,self.kids)
        self.meta_feats=meta_feats[mask]

    def _generate_train_(self):
        if self.ifTrain:
            # np.memmap(file_path, dtype="float32", mode="r")
            all_datas=np.load(os.path.join(self.npy_dir_path,"all_datas_Tr.npy"),mmap_mode="r+")
            # all_datas=np.load(os.path.join(self.npy_dir_path,"all_datas_Tr.npy"))
            all_labels=np.load(os.path.join(self.npy_dir_path,"all_labels_Tr.npy"))
            all_kids=np.load(os.path.join(self.npy_dir_path,"all_kids_Tr.npy"))
            p2n_rate_data=self.p2n_rate_train
        else:
            all_datas=np.load(os.path.join(self.npy_dir_path,"all_datas_t.npy"),mmap_mode="r+")
            # all_datas=np.load(os.path.join(self.npy_dir_path,"all_datas_t.npy"))
            all_labels=np.load(os.path.join(self.npy_dir_path,"all_labels_t.npy"))
            all_kids=np.load(os.path.join(self.npy_dir_path,"all_kids_t.npy"))
            p2n_rate_data=self.p2n_rate_test
        self.num_wo_add=len(all_kids)

        if p2n_rate_data>0:
            mask=all_labels==1
            num_1=np.sum(mask)
            num_keep_0=int(num_1/p2n_rate_data)
            idxs_0=np.argwhere(~mask).flatten()
            np.random.seed(123)
            idxs_0=np.random.permutation(idxs_0)[:num_keep_0]
            mask[idxs_0]=True
            self.use_iiis=np.argwhere(mask).flatten()
        else:
            self.use_iiis=np.arange(len(all_labels))

        self.all_datas=all_datas
        self.all_labels=all_labels
        self.all_kids=all_kids

    def __len__(self):
        return len(self.use_iiis)

    def __getitem__(self, index):
        index=self.use_iiis[index]
        label=self.all_labels[index]
        lc_time,lc_flux,lc_label=self.all_datas[index]
        lc_id=self.all_kids[index]

        lc_mask=np.isnan(lc_flux)
        lc_flux[lc_mask]=1
        lc_mask=lc_mask.astype(np.float_)

        meta_feats=self.meta_feats[lc_id]
        kid_name=self.kids[lc_id]

        idx_noNan=np.argwhere(~np.isnan(lc_time)).flatten()[0]
        value_noNan=lc_time[idx_noNan]
        idxs_Nan=np.argwhere(np.isnan(lc_time)).flatten()
        values_Nan=np.ones_like(idxs_Nan)*value_noNan+(idxs_Nan-idx_noNan)*(1/48)
        lc_time[idxs_Nan]=values_Nan

        return lc_time,lc_flux,lc_label,lc_mask,label,meta_feats,kid_name,lc_id


# Adapted from: https://github.com/ufoym/imbalanced-dataset-sampler/blob/master/torchsampler/imbalanced.py
# Copyright (c) 2018 Ming
# License: MIT
class ImbalancedDatasetSampler(torch.utils.data.sampler.Sampler):
    """Samples elements randomly from a given list of indices for imbalanced dataset

    Arguments:
        indices: a list of indices
        num_samples: number of samples to draw
        callback_get_label: a callback-like function which takes two arguments - dataset and index
    """

    def __init__(
        self,
        dataset,
        labels: list = None,
        indices: list = None,
        num_samples: int = None,
        callback_get_label: Callable = None,
    ):
        # if indices is not provided, all elements in the dataset will be considered
        self.indices = list(range(len(dataset))) if indices is None else indices

        # define custom callback
        self.callback_get_label = callback_get_label

        # if num_samples is not provided, draw `len(indices)` samples in each iteration
        self.num_samples = len(self.indices) if num_samples is None else num_samples

        # distribution of classes in the dataset
        df = pd.DataFrame()
        df["label"] = self._get_labels(dataset) if labels is None else labels
        df.index = self.indices
        df = df.sort_index()

        label_to_count = df["label"].value_counts()

        weights = 1.0 / label_to_count[df["label"]]

        self.weights = torch.DoubleTensor(weights.to_list())

    def _get_labels(self, dataset):
        if isinstance(dataset, torch.utils.data.Dataset):
            # return dataset.use_label[dataset.use_iiis]
            return dataset.all_labels[dataset.use_iiis]
        else:
            raise NotImplementedError

    def __iter__(self):
        return (self.indices[i] for i in torch.multinomial(self.weights, self.num_samples, replacement=True))

    def __len__(self):
        return self.num_samples