Python LSTMNet示例

示例#1

文件： LSTM_REFLECTION_TIME_testing.py 项目： savagewil/GoQLearning

from MLLibrary.Models.LSTMNet import LSTMNet
from random import choice
import numpy
import matplotlib.pyplot as plt
from datetime import datetime

S = 1
LSTM = LSTMNet(S, S, S + 1)
Iterations = 20000
learning_rate = 0.5
start_time = datetime.now()

inputs = []
error = []
accuracy = []
in_ = []
label = []
prediction = []
expected = []

input = [[]]
for j in range(S):
    input[0].append(choice([0.0, 1.0]))  #-1.0,
inputs.append(input)

for i in range(Iterations):
    input = [[]]
    for j in range(S):
        input[0].append(choice([0.0, 1.0]))  #-1.0,
    inputs.append(input)

示例#2

import sys

from MLLibrary.Models.LSTMNet import LSTMNet
import numpy

I = 1
O = 4
D = 4
LSTM = LSTMNet(I, O, D)
MAX_ITER = 1000000
BATCH = 100
LEARNING_RATIO = 3.0


def get_X():
    index = 1
    while True:
        index += 1
        numpy.random.seed(index)
        yield [[numpy.random.choice([0.0, 1.0])]]


def get_Y():
    index = 1
    while True:
        numpy.random.seed(index)
        b1 = numpy.random.choice([0.0, 1.0])
        index += 1
        numpy.random.seed(index)
        b2 = numpy.random.choice([0.0, 1.0])
        yield [[(b1 == 1.0 and b2 == 1.0), None, None, None]]

示例#3

文件： LSTM_AND_testing.py 项目： savagewil/GoQLearning

from MLLibrary.Models.LSTMNet import LSTMNet
from random import choice
import numpy
import matplotlib.pyplot as plt

I = 2
O = 1
D = 2
LSTM = LSTMNet(I, O, D)

inputs = []
error = []
accuracy = []
in_ = []
label = []
for i in range(10000):
    input = [[]]
    for j in range(I):
        input[0].append(choice([0.0, 1.0])) #-1.0,
    inputs.append(input)

    if len(inputs) > 2:
        inputs = inputs[-2:]
    X = numpy.array(input)
    Y = numpy.array(1.0 if (input[0][0] == 1.0 and input[0][0] == 1.0) else 0.0) #inputs[0]
    LSTM.set_in(X)
    P = LSTM.get_out()
    d = numpy.zeros((1, 3))
    d = (Y - P)/2.0
    error.append(numpy.sum((Y - P)**2.0))
    accuracy.append(numpy.sum((Y - numpy.round(P))**2.0))

示例#4

from MLLibrary.Models.LSTMNet import LSTMNet
from random import choice
import numpy
import matplotlib.pyplot as plt

S = 1
LSTM = LSTMNet(S, S, S + 1)

inputs = []
error = []
accuracy = []
in_ = []
label = []
for i in range(10000):
    input = [[]]
    for j in range(S):
        input[0].append(choice([0.0, 1.0]))  #-1.0,
    inputs.append(input)

    if len(inputs) > 2:
        inputs = inputs[-2:]
    X = numpy.array(input)
    Y = numpy.array(input)  #inputs[0]
    LSTM.set_in(X)
    P = LSTM.get_out()
    d = numpy.zeros((1, 3))
    d = (Y - P) / 2.0
    error.append(numpy.sum((Y - P)**2.0))
    accuracy.append(numpy.sum((Y - numpy.round(P))**2.0))
    in_.append(input[0][0])
    label.append("One" if (input[0][0] == 1.0) else "Zero")

示例#5

文件： LSTM_AND_batch_testing.py 项目： savagewil/GoQLearning

import sys

from MLLibrary.Models.LSTMNet import LSTMNet
import matplotlib.pyplot as plt

from MLLibrary.StatsHandler import StatsHandler

I = 2
O = 1
D = 2

statsHandler = StatsHandler()

LSTM = LSTMNet(I, O, D, statsHandler=statsHandler)
MAX_ITER = 1000
BATCH = 10
LEARNING_RATIO = 0.01

def get_X():
    while True:
        for b1 in [0.0, 1.0]:
            for b2 in [0.0, 1.0]:
                yield [[b1, b2]]

def get_Y():
    while True:
        for b1 in [False, True]:
            for b2 in [False, True]:
                yield 1.0 if (b1 and b2) else 0.0