import pickle from sarcos import download_sarcos from sarcos import load_sarcos from sarcos import nMSE from sklearn.preprocessing import StandardScaler from multilayer_neural_network import MultilayerNeuralNetwork from minibatch_sgd import MiniBatchSGD if __name__ == "__main__": np.random.seed(0) # Download Sarcos dataset if this is required download_sarcos() # Load training set and test set X, Y = load_sarcos("train") X_test, Y_test = load_sarcos("test") # Scale targets target_scaler = StandardScaler() Y = target_scaler.fit_transform(Y) Y_test = target_scaler.transform(Y_test) # Train model (code for exercise 10.2 1/2/3) ############################################################################ # Train neural network D = (X.shape[1], ) F = Y.shape[1] layers = \ [ { "type": "fully_connected",
] sorted_test_error = [ error for _, error in sorted(zip(indices, test_error_list)) ] data = { 'params': sorted_param, 'train_error': sorted_train_error, 'test_error': sorted_test_error } df = pd.DataFrame(data=data) df.to_csv(path, index=False) if __name__ == "__main__": download_sarcos() X_train, Y_train = load_sarcos('train') X_train, Y_train = X_train[0:30000], Y_train[0:30000] target_scaler, feature_scaler = StandardScaler(), StandardScaler() Y_train, X_train = target_scaler.fit_transform( Y_train), feature_scaler.fit_transform(X_train) layers = [{"type": "fully_connected", "num_nodes": 90}] parameters = { 'alpha': [0.0005, 0.003, 0.01], 'alpha_decay': [0.95, 0.97, 1], 'batch_size': [30, 55, 80], 'eta': [0.2, 0.5, 0.8], 'eta_inc': [0, 0.00001] } scores, params = [], [] F = Y_train.shape[1] D = (X_train.shape[1], )
from sarcos import download_sarcos from sarcos import load_sarcos from sarcos import nMSE from sklearn.preprocessing import StandardScaler from multilayer_neural_network import MultilayerNeuralNetwork from minibatch_sgd import MiniBatchSGD import time import numpy as np if __name__ == "__main__": download_sarcos() X_train, Y_train = load_sarcos('train') X_test, Y_test = load_sarcos('test') target_scaler, feature_scaler = StandardScaler(), StandardScaler() Y_train, X_train = target_scaler.fit_transform( Y_train), feature_scaler.fit_transform(X_train) Y_test, X_test = target_scaler.transform(Y_test), feature_scaler.transform( X_test) layers = [{"type": "fully_connected", "num_nodes": 90}] F = Y_train.shape[1] D = (X_train.shape[1], ) model = MultilayerNeuralNetwork(D, F, layers, training='regression', std_dev=0.01, verbose=True) mbsgd = MiniBatchSGD(net=model, epochs=100,