from scipy.spatial.distance import euclidean from fastdtw import fastdtw from data import character_trajectories import os import numpy as np from keras.models import load_model import matplotlib.pyplot as plt from rnn import create_rnn from minisom import MiniSom DIR_PATH = os.path.dirname(os.path.realpath(__file__)) DIR_MODEL = os.path.join(DIR_PATH, "rnn_models") RNN_FILE_MODEL = os.path.join(DIR_MODEL, "model_RNN_205_4_noise_04_v0.hdf5") """reshape in and outputs""" (X_train, Y_train), (X_test, Y_test) = character_trajectories.load_data( '../RNN/data/char_trajectories_4.pkl') print("x_train shape: ", X_train.shape) print("y_train shape: ", Y_train.shape) print("x_test shape: ", X_test.shape) print("y_test shape: ", Y_test.shape) """load model""" rnn = create_rnn(clip=True, noise=True, std_dev=0, hidden_size=205, double_stacked=False, input_shape=(X_train.shape[1], X_train.shape[2])) rnn.load_weights(RNN_FILE_MODEL) def pad_class(class_vector, x=1):
from data import character_trajectories import os import numpy as np from keras.models import load_model import matplotlib.pyplot as plt DIR_PATH = os.path.dirname(os.path.realpath(__file__)) DIR_MODEL = os.path.join(DIR_PATH, "rnn_models") dim = 20 types = False """reshape in and outputs""" (X_train, Y_train), (X_test, Y_test) = character_trajectories.load_data( '../RNN/data/char_trajectories_{}.pkl'.format(dim)) """remove the padding of the x_data""" X_train = X_train[:, 0, :] X_test = X_test[:, 0, :] if types: """reduce the types to characters by taking the max along the axis""" print("test sample before the type reduction: ", X_test[0]) X_train = np.amax(X_train, axis=2) X_test = np.amax(X_test, axis=2) print("test sample after type reduction: ", print(X_test[0])) print("x_train shape: ", X_train.shape) print("y_train shape: ", Y_train.shape) print("x_test shape: ", X_test.shape) print("y_test shape: ", Y_test.shape) for i in range(5):
import os import numpy as np import keras.backend as K DIR_PATH = os.path.dirname(os.path.realpath(__file__)) DIR_MODEL = os.path.join(DIR_PATH, "rnn_models") #RNN_FILE_MODEL = os.path.join(DIR_MODEL, "model_RNN_205_2_naked_v0.hdf5") print(K.tensorflow_backend._get_available_gpus()) types = False dim = 20 batch_size = 1 sparse = False sparse_val = True # load training and validation data #x4_types_centers_only (X_train, Y_train), (X_test, Y_test) = character_trajectories.load_data( '../RNN/data/char_trajectories_{}_adverserial.pkl'.format(dim)) (X_val, Y_val), (_, _) = character_trajectories.load_data( '../RNN/data/char_trajectories_{}x1_types_centers_only.pkl'.format(dim)) if sparse: """reduce the types to characters by taking the max along the axis""" print("test sample before the type reduction: ", X_train[0]) X_train = np.amax(X_train, axis=3) X_test = np.amax(X_test, axis=3) print("test sample after type reduction: ", print(X_train[0])) if sparse_val: #reshape validation set X_val = np.amax(X_val, axis=3) print("x_train shape: ", X_train.shape)
from keras.models import Model, Sequential from keras.layers import Input, LSTM, Dense, Flatten from data import character_trajectories from keras.callbacks import ModelCheckpoint import os import numpy as np DIR_PATH = os.path.dirname(os.path.realpath(__file__)) DIR_MODEL = os.path.join(DIR_PATH, "rnn_models") #RNN_FILE_MODEL_weights = os.path.join(DIR_MODEL, "weights_RNN_v1.hdf5") types = False sparse = True dim = 20 # load training and validation data (x_train, y_train), (x_test, y_test) = character_trajectories.load_data( "../RNN/data/char_trajectories_{}x1_types_centers_only.pkl".format(dim)) """remove the padding of the x_data""" x_train = x_train[:, 0, :] x_test = x_test[:, 0, :] if types or sparse: """reduce the types to characters by taking the max along the axis""" print("test sample before the type reduction: ", x_train[0]) x_train = np.amax(x_train, axis=2) x_test = np.amax(x_test, axis=2) print("test sample after type reduction: ", print(x_train[0])) print("x_train shape: ", x_train.shape) print("y_train shape: ", y_train.shape) print("x_test shape: ", x_test.shape) print("y_test shape: ", y_test.shape)
from data import character_trajectories import os import numpy as np from keras.models import load_model from sklearn.preprocessing import normalize import matplotlib.pyplot as plt DIR_PATH = os.path.dirname(os.path.realpath(__file__)) DIR_MODEL = os.path.join(DIR_PATH, "rnn_models") RNN_FILE_MODEL = os.path.join(DIR_MODEL, "model_RNN_one_shot_std.hdf5") # load training and validation data (x_train, y_train), (x_val, y_val), (x_test, y_test) = character_trajectories.load_data() print("x_train/val/test shape: " ,x_train.shape) print("y_train/val/test shape: " ,y_train.shape) """reshape data""" x_train = np.reshape(x_train, (len(x_train), 1, 20)) y_train = np.reshape(y_train, (len(y_train), 205, 2)) x_test = np.reshape(x_test, (len(x_test), 1, 20)) y_test = np.reshape(y_test, (len(y_test), 2, 205)) """load model""" rnn = load_model(RNN_FILE_MODEL)