5 CL3

import numpy as np

from sklearn.model_selection import train_test_split

from sklearn.preprocessing import StandardScaler

from sklearn.neural_network import MLPRegressor

from geneticalgorithm import geneticalgorithm as ga

import warnings



def fitness_function(params):

    

    hidden_layer_sizes = (int(params[0]),) * int(params[1])

    activation = ['identity', 'logistic', 'tanh', 'relu'][int(params[2])]

    solver = ['lbfgs', 'sgd', 'adam'][int(params[3])]

    

   

    model = MLPRegressor(hidden_layer_sizes=hidden_layer_sizes, activation=activation, solver=solver)

    model.fit(X_train, y_train)

    

    

    fitness = -model.score(X_val, y_val)

    

    return fitness


def train_neural_network(params):

    

    hidden_layer_sizes = (int(params[0]),) * int(params[1])

    activation = ['identity', 'logistic', 'tanh', 'relu'][int(params[2])]

    solver = ['lbfgs', 'sgd', 'adam'][int(params[3])]

    

    

    model = MLPRegressor(hidden_layer_sizes=hidden_layer_sizes, activation=activation, solver=solver)

    model.fit(X_train, y_train)

    

    

    validation_error = -model.score(X_val, y_val)

    

    return validation_error



np.random.seed(42)



num_samples = 1000

num_features = 5  



X = np.random.rand(num_samples, num_features)



true_coefficients = np.random.rand(num_features) * 10  

noise = np.random.normal(loc=0, scale=1, size=num_samples)  

y = np.dot(X, true_coefficients) + noise




X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2, random_state=42)


warnings.filterwarnings("ignore")



number_of_parameters = 4  

varbound = np.array([[5, 50],  

                     [1, 5],    

                     [0, 3],    

                     [0, 2]])   

algorithm_param = {'max_num_iteration': 1, 'population_size': 100, 'mutation_probability': 0.1,

                   'elit_ratio': 0.01, 'crossover_probability': 0.5, 'parents_portion': 0.3,

                   'crossover_type': 'uniform', 'max_iteration_without_improv': None}

model = ga(function=fitness_function, dimension=number_of_parameters, variable_type='int', variable_boundaries=varbound,

           algorithm_parameters=algorithm_param)



model.run()



best_params = model.output_dict['variable']



validation_error = train_neural_network(best_params)


print("Best parameters found by genetic algorithm:", best_params)

print("Validation error of neural network:", validation_error)



Comments

Post a Comment

Popular posts from this blog

DREAM LEAGUE SOCCER 2019 MOD

Image
HEY GUYS THIS IS THE DLS19 MOD APK AND OBB FILE  MOD INCLUDE - UNLIMITED MONEY                               UNLOCKED ALL PLAYERS                                ADS FREE Dream League Soccer 2019  Dream League Soccer 2016 is here, and it’s better than ever! Soccer as we know it has changed, and this is YOUR chance to build THE best team on the planet. Recruit real FIFPro™ licensed superstars, build your own stadium, and take on the World with Dream League Online as you march towards glory, on your road to Soccer Stardom! Download Dream League Soccer 2016 for FREE now! * Includes tablet support! * ***************************************************** IMPORTANT * This game is free to play, but additional content and in-game items may be purchased for real money. * Coins can be earned during game...
Read more

10 CL3

 import numpy as np import random def distance(city1, city2):     return np.linalg.norm(city1 - city2)  num_cities = 4 num_ants = 4 evaporation_rate = 0.05 alpha = 1 beta = 1 cities = np.random.rand(num_cities, 2) distances = [] for city1 in cities:     temp = []     for city2 in cities:          temp.append(distance(city1, city2))     distances.append(temp) phermones = np.random.rand(num_cities, num_cities) tour_lengths = [] tour_paths = [] for _ in range(num_ants):     randomly_selected_city = random.randint(0, num_cities-1)     selected_city = randomly_selected_city     tour_length = 0     tour_path = []     for j in range(num_cities):          tour_path.append(selected_city)         dist = distances[selected_city]         phero = phermones[selected_city]         stochastic_gra...
Read more

Garena Contra Return

Image
Pre-register now to receive Diamond *20, Gold *1888, and Team Revival Coin *2! Remember the secret code, Up Up Down Down Left Right Left Right BA? Contra is back! Together with KONAMI and Tencent Timi Studio Group, Garena has released the classic sequel, “Contra: Return!” “Contra: Return” retains the side-scrolling shooter action gameplay, along with duo mode and additional new features. Defeat ultimate bosses and relive arcade glories. With new action packed gameplay and HD high-definition graphics, 3D character, vivid sound and animation effects, you are in for an extraordinary experience! Unleash the warrior in you! Choose from an arsenal of heroes and weapons and challenge yourself in PVP instant battles, team modes and face our ultimate challenge, Underwater 8! Join our legendary heroes Bill and Lance in their fight against menacing aliens and together, we shall save the world! – Challenge a New Classic Licensed by KONAMI, the classic sequel Contra: Return is bac...
Read more