X_train, Y_train = X, Y
Y_vec = Utils.vectorize_output(Y_train, nb_labels)
print('Epoch %d' % (i+1))
print("-"*30)
nn.initialize_weights()
start_time = timer()
j_history = nn.train(X_train, Y_vec, alpha, lbda, momentum, precision, nb_iters, verbose=False)
total_time = timer() - start_time
a.append(len(j_history))
b.append(total_time)
c.append(Statistics.accuracy(nn.predict_classes(X_train), Y_train))
d.append(np.mean(np.array(j_history)))
# print('Iterations:\t %d' % len(j_history))
# print('Time:\t\t %.9f seconds' % total_time)
# print('Accuracy train:\t %.2f' % Statistics.accuracy(nn.predict_classes(X_train), Y_train))
# print('\n')
# d.append(np.mean(np.array(j_history)))
# colors = "gbrcmyk"*2
# Plots.lineplot(list(range(len(j_history))), j_history, color=colors[k], label='Alpha {}'.format(alpha))
Plots.draw_boundary_rnn(X, nn.predict_classes, nb_hidden=nb_hidden)
# print('Epoch %d' % (i+1))
# print("-"*30)
sys.stdout.write("Degree %2d - Alpha %2.2lf - Epoch %2d \r" % (nb_degree, alpha, i+1))
sys.stdout.flush()
start_time = timer()
lg.initialize_weights()
j_history = lg.train(X, Y, alpha, lbda, precision, nb_iters, verbose=False)
total_time = timer() - start_time
a.append(len(j_history))
b.append(total_time)
c.append(Statistics.accuracy(lg.predict_binary(X), Y))
d = j_history
# print('Iterations:\t %d' % a[-1])
# print('Time:\t\t %.2f seconds' % b[-1])
# print('Accuracy:\t %.2f' % c[-1])
# f1, precision, recall, acc = Statistics.f_score(lg.predict_binary(X), Y)
# print('F1:\t %.2f' % f1)
# print('Prec:\t %.2f' % precision)
# print('Rec:\t %.2f' % recall)
# print('Acc:\t %.2f' % acc)
# print('\n')
# if nb_degree == 2:
sys.stdout.flush()
start_time = timer()
lg.initialize_weights()
j_history = lg.train(X,
Y,
alpha,
lbda,
precision,
nb_iters,
verbose=False)
total_time = timer() - start_time
a.append(len(j_history))
b.append(total_time)
c.append(Statistics.accuracy(lg.predict_binary(X), Y))
d = j_history
# print('Iterations:\t %d' % a[-1])
# print('Time:\t\t %.2f seconds' % b[-1])
# print('Accuracy:\t %.2f' % c[-1])
# f1, precision, recall, acc = Statistics.f_score(lg.predict_binary(X), Y)
# print('F1:\t %.2f' % f1)
# print('Prec:\t %.2f' % precision)
# print('Rec:\t %.2f' % recall)
# print('Acc:\t %.2f' % acc)
# print('\n')
# if nb_degree == 2:
# move enemies and projectiles
moveEnemies()
moveProjectiles()
# check for any collisions
if collisionDetection(DISPLAYSURF, BLACK, enemy.getEnemyList(), proj.projectile_list, my_character) == True:
# add player to the database
insertPlayerRecord(statistics.name, statistics.total_kills, statistics.total_deaths, statistics.blue_kills, statistics.green_kills, statistics.red_kills, statistics.purple_kills, statistics.shots_fired, statistics.shots_hit, statistics.accuracy)
# clear the display and show the play again menu
DISPLAYSURF.fill((0,0,0))
playAgainMenuOptions = [Option(DISPLAYSURF, BLUE, font, "YES", (config.display_x // 2 - 30, config.display_y // 2)),
Option(DISPLAYSURF, BLUE, font, "NO", (config.display_x // 2 - 20, config.display_y // 2 + 100)), Option(DISPLAYSURF, BLUE, font, "MAIN MENU", (config.display_x // 2 - len("MAIN MENU") * 10, config.display_y - 100))]
while(playAgainMenu):
playAgainMenu = drawPlayAgainMenu(DISPLAYSURF)
# update statistics
if statistics.shots_fired != 0:
statistics.accuracy = statistics.shots_hit / statistics.shots_fired
statistics.accuracy = round(statistics.accuracy, 2)
# slowly step the difficulty up
config.difficulty += config.difficulty_scaler
# draw the score onto the surface
scoreText = font.render("Score: " + str(statistics.total_kills), True, WHITE, BLACK)
textRect = scoreText.get_rect()
DISPLAYSURF.blit(scoreText, textRect)
pygame.display.update()
nn.initialize_weights()
start_time = timer()
j_history = nn.train(X_train,
Y_vec,
alpha,
lbda,
momentum,
precision,
nb_iters,
verbose=False)
total_time = timer() - start_time
a.append(len(j_history))
b.append(total_time)
c.append(Statistics.accuracy(nn.predict_classes(X_train), Y_train))
d.append(np.mean(np.array(j_history)))
# print('Iterations:\t %d' % len(j_history))
# print('Time:\t\t %.9f seconds' % total_time)
# print('Accuracy train:\t %.2f' % Statistics.accuracy(nn.predict_classes(X_train), Y_train))
# print('\n')
# d.append(np.mean(np.array(j_history)))
# colors = "gbrcmyk"*2
# Plots.lineplot(list(range(len(j_history))), j_history, color=colors[k], label='Alpha {}'.format(alpha))
Plots.draw_boundary_rnn(X, nn.predict_classes, nb_hidden=nb_hidden)
ma, da = np.mean(np.array(a)), np.std(np.array(a))
