search

lukasruff / Deep-SVDD-PyTorch

A PyTorch implementation of the Deep SVDD anomaly detection method
MIT License
698 stars 197 forks source link

Pre-computed min-max values #12

Closed kumarneelabh13 closed 5 years ago

kumarneelabh13 commented 5 years ago

I think there's a problem with precomputing min-max values for separate classes. The problem is that this technique will not be suitable for a new and "unknown" sample.

from datasets/cifar10.py

    # Pre-computed min and max values (after applying GCN) from train data per class

min_max = [(-28.94083453598571, 13.802961825439636), (-6.681770233365245, 9.158067708230273), (-34.924463588638204, 14.419298165027628), (-10.599172931391799, 11.093187820377565), (-11.945022995801637, 10.628045447867583), (-9.691969487694928, 8.948326776180823), (-9.174940012342555, 13.847014686472365), (-6.876682005899029, 12.282371383343161), (-15.603507135507172, 15.2464923804279), (-6.132882973622672, 8.046098172351265)]

Or am I missing something?

kumarneelabh13 commented 5 years ago

I got it. So all the train and test samples get re-scaled according to the normal class values.

matteoguarrera commented 3 years ago 
train_set_full = MyMNIST(root=root, train=True, download=True,
                        transform=None, target_transform=None)

MIN = []
MAX = []
for normal_classes in range(10):
    train_idx_normal = get_target_label_idx(train_set_full.train_labels.clone().data.cpu().numpy(), normal_classes)
    train_set = Subset(train_set_full, train_idx_normal)

    _min_ = []
    _max_ = []
    for idx in train_set.indices:
        gcm = global_contrast_normalization(train_set.dataset.data[idx].float(), 'l1')
        _min_.append(gcm.min())
        _max_.append(gcm.max())
    MIN.append(np.min(_min_))
    MAX.append(np.max(_max_))
print(list(zip(MIN, MAX)))