nbof.py 文件源码

def initialize_dictionary(self, X, max_iter=100, redo=5, n_samples=50000, normalize=False):
        """
        Samples some feature vectors from X and learns an initial dictionary
        :param X: list of objects
        :param max_iter: maximum k-means iters
        :param redo: number of times to repeat k-means clustering
        :param n_samples: number of feature vectors to sample from the objects
        :param normalize: use l_2 norm normalization for the feature vectors
        """

        # Sample only a small number of feature vectors from each object
        samples_per_object = int(np.ceil(n_samples / len(X)))

        features = None
        print("Sampling feature vectors...")
        for i in (range(len(X))):
            idx = np.random.permutation(X[i].shape[0])[:samples_per_object + 1]
            cur_features = X[i][idx, :]
            if features is None:
                features = cur_features
            else:
                features = np.vstack((features, cur_features))

        print("Clustering feature vectors...")
        features = np.float64(features)
        if normalize:
            features = feature_normalizer(features)

        V = cluster.k_means(features, n_clusters=self.Nk, max_iter=max_iter, n_init=redo)
        self.V.set_value(np.asarray(V[0], dtype=theano.config.floatX))