pyanomaly.core.other package¶
Submodules¶
pyanomaly.core.other.kmeans module¶
@author: Yuhao Cheng @contact: yuhao.cheng[at]outlook.com
-
pyanomaly.core.other.kmeans.initialize(X, num_clusters)¶ initialize cluster centers :param X: (torch.tensor) matrix :param num_clusters: (int) number of clusters :return: (np.array) initial state
-
pyanomaly.core.other.kmeans.kmeans(X, num_clusters, distance='euclidean', cluster_centers=[], tol=0.0001, device=device(type='cpu'))¶ perform kmeans :param X: (torch.tensor) matrix :param num_clusters: (int) number of clusters :param distance: (str) distance [options: 'euclidean', 'cosine'] [default: 'euclidean'] :param tol: (float) threshold [default: 0.0001] :param device: (torch.device) device [default: cpu] :return: (torch.tensor, torch.tensor) cluster ids, cluster centers
-
pyanomaly.core.other.kmeans.kmeans_predict(X, cluster_centers, distance='euclidean', device=device(type='cpu'))¶ predict using cluster centers :param X: (torch.tensor) matrix :param cluster_centers: (torch.tensor) cluster centers :param distance: (str) distance [options: 'euclidean', 'cosine'] [default: 'euclidean'] :param device: (torch.device) device [default: 'cpu'] :return: (torch.tensor) cluster ids
-
pyanomaly.core.other.kmeans.pairwise_cosine(data1, data2, device=device(type='cpu'))¶
-
pyanomaly.core.other.kmeans.pairwise_distance(data1, data2, device=device(type='cpu'))¶