display = ShowTraining(epochs_num=epochs)
trainer = Trainer(show_training=True, show_function=display.show)
t = time.time()
J_train_list, dJdy_list, J_test_list = trainer.learn(
model=model,
train=train,
test=test,
loss=NegativeLogLikelihoodLoss(),
# loss = CrossEntropyLoss(),
# loss = SquaredLoss(),
# optimizer = GradientDescent(learning_rate = 0.15/110),
optimizer=GradientDescentMomentum(learning_rate=0.005, momentum=0.8),
batch_size=100,
epochs=epochs)
elapsed = time.time() - t
print 'Training time: ' + str(elapsed)
y1 = []
for i, (x, target) in enumerate(train):
y1.append(model.forward(x))
#
y2 = []
for i, (x, target) in enumerate(test):
y2.append(model.forward(x))
#
np.array([-1.0, -1.0]), np.array([1.0, 1.0]))
n = Sequential(
norm,
# LinearLayer(2,5,weights='gaussian'),
# TanhLayer,
#AddGaussian(1),
LinearLayer(2, 4, weights='gaussian'),
RandomGaussianLayer(1),
SoftMaxLayer)
agent = GenericAgent(n, 4, 40, 5.0)
agent.set_training_options(
Trainer(),
NegativeLogLikelihoodLoss(),
GradientDescentMomentum(
learning_rate=0.1,
momentum=0.7) #GradientDescent(learning_rate=0.2)
)
start = np.array([3.5, 3.5])
obstacles = [
# np.array([2.5,2.5,1.0])
]
win = np.array([0.5, 0.5, 0.5])
def data_gen(t=0):
ball = Ball(np.random.rand(1, 2)[0] * 5)
time_start_game = 0
for ind, time in enumerate(np.linspace(0, time_end, time_end / time_step)):
state = np.array([ball.p[0], ball.p[1]])
# W3 = utils.SharedWeights(np.array([[10.0,-10.0,0.0],[-10.0,10.0,0.0]]),2+1,2)
#W2 = utils.SharedWeights('gaussian',2+1,2)
Q_hat = Sequential(
norm,
LinearLayer(2, 2, weights=W3),
ReluLayer,
LinearLayer(2, 3, weights=W4),
# TanhLayer
)
#Q, Q_hat, replay_memory_size, minibatch_size = 100, learning_rate = 0.1, gamma = 0.95, policy = 'esp-greedy', epsilon = 0.3
agent = DeepAgent(Q, Q_hat, 1000, minibatch_size=100, policy='eps-greedy')
agent.set_training_options(
Trainer(show_training=True),
SquaredLoss(),
# GradientDescent(learning_rate=0.001)
GradientDescentMomentum(learning_rate=0.1, momentum=0.2, clip=1) #
)
J_train_list = []
dJdy_list = []
def data_gen(t=0):
cart = Cart()
ball = Ball()
catches = 0
for ind, time in enumerate(np.linspace(0, time_end, time_end / time_step)):
# print time
state = np.array([ball.p[0], cart.p[0]]) #,ball.p[1],ball.v[0]])
ind_command = agent.forward(state)
printer = ShowTraining(epochs_num=epochs,
weights_list={
'a': a.net.W,
'b': b.net.W,
'c': c.net.W
})
t = Trainer(show_training=True, show_function=printer.show)
#
J_list, dJdy_list, J_test_list = t.learn(
model=n,
train=train,
test=test,
loss=SquaredLoss(),
optimizer=GradientDescentMomentum(learning_rate=0.9, momentum=0.5),
# batch_size = len(train),
# optimizer = AdaGrad(learning_rate=0.6),
epochs=epochs)
#
plt.figure(4)
plt.title('Errors History (J)')
plt.plot(np.array([x for (x, t) in test]),
np.array([t for (x, t) in test]),
color='red')
plt.plot(np.array([x for (x, t) in test]),
np.array([n.forward(x) for (x, t) in test]),
color='green')
# plt.ylim([0, 2])
plt.xlabel('x')