def init_weights_randomly(self, vmin= -0.25, vmax=0.25):
'''
initialize the perceptron with random weights between [vmin ; vmax]
returns None
'''
self.weights = [randmm(vmin, vmax) for _ in range(self.nbr_input)]
if(self._enable_bias):
self.bias = randmm(vmin, vmax)
self._last_weights = list(self.weights) #_last_weights is used by the momentum algo
self._last_bias = self.bias
self._weights_updated = True
def init_weights_randomly(self, vmin=-0.25, vmax=0.25):
'''
initialize the perceptron with random weights between [vmin ; vmax]
returns None
'''
self.weights = [randmm(vmin, vmax) for _ in range(self.nbr_input)]
if (self._enable_bias):
self.bias = randmm(vmin, vmax)
self._last_weights = list(
self.weights) #_last_weights is used by the momentum algo
self._last_bias = self.bias
self._weights_updated = True
def _line_to_list(self, line):
lstring = list(line.replace(" ", "").replace("\n", ""))
result = []
for sym in lstring:
for key in self._rules:
if(sym == key):
if(len(self._rules[key]) == 1):
result += [self._rules[key][0]]
else:
result += [randmm(self._rules[key][0], self._rules[key][1])]
return result
def _line_to_list(self, line):
lstring = list(line.replace(" ", "").replace("\n", ""))
result = []
for sym in lstring:
for key in self._rules:
if (sym == key):
if (len(self._rules[key]) == 1):
result += [self._rules[key][0]]
else:
result += [
randmm(self._rules[key][0], self._rules[key][1])
]
return result
def __init__(self, control):
self.nbr_input = control.nbr_input
self.learning_rate = control.learning_rate
self.momentum = control.momentum
self.temperature = control.temperature
self._ntype = control._ntype
self._enable_bias = control._enable_bias
self.weights = []
for w in control.weights:
self.weights.append(w)
for _ in range(20):
self.weights.append(randmm(-1, 1))
self.bias = control.bias
self._last_bias = control._last_bias
self._last_weights = list(self.weights)
self._weights_updated = True
self._last_inputs = []
def __init__(self, control):
self.nbr_input = control.nbr_input
self.learning_rate = control.learning_rate
self.momentum = control.momentum
self.temperature = control.temperature
self._ntype = control._ntype
self._enable_bias = control._enable_bias
self.weights = []
for w in control.weights:
self.weights.append(w)
for _ in range(20):
self.weights.append(randmm(-1, 1))
self.bias = control.bias
self._last_bias = control._last_bias
self._last_weights = list(self.weights)
self._weights_updated = True
self._last_inputs = []
def init_weights_randomly(self, vmin=-0.25, vmax=0.25):
self.weights = [randmm(vmin, vmax) for _ in range(self.nbr_input)]
self.bias = randmm(vmin, vmax)
ho.init_weights_randomly(-1, 1)
high_order.append(ho)
#trials
wins = []
wagers = []
last_output = [ [0 for _ in range(5)] for _ in range(nbr_network)]
for epoch in range(nbr_epoch):
nbr_win = 0
nbr_gwager = 0
for net in range(nbr_network):
for trial in range(nbr_trial):
for i in range(len(last_output[net])):
last_output[net][i] += randmm(0.01, 0.03)
bet = index_max(first_order[net].calc_output(last_output[net]))
for i in range(len(first_order[net].stateHiddenNeurons)):
first_order[net].stateHiddenNeurons[i] += randmm(0.01, 0.03)
wager = index_max(high_order[net].calc_output(first_order[net].stateHiddenNeurons))
win = deck(bet)
if((win[0] and wager == 0)):
nbr_gwager += 1
if(win[0]):
high_order[net].train(first_order[net].stateHiddenNeurons, [1. , grid])
def init_weights_randomly(self, vmin= -0.25, vmax=0.25):
self.weights = [randmm(vmin, vmax) for _ in range(self.nbr_input)]
self.bias = randmm(vmin, vmax)
ho = MultilayerPerceptron(40, 40, 2, grid, 0.0003, 0., 1., False, True)
ho.init_weights_randomly(-1, 1)
high_order.append(ho)
#trials
wins = []
wagers = []
last_output = [[0 for _ in range(5)] for _ in range(nbr_network)]
for epoch in range(nbr_epoch):
nbr_win = 0
nbr_gwager = 0
for net in range(nbr_network):
for trial in range(nbr_trial):
for i in range(len(last_output[net])):
last_output[net][i] += randmm(0.01, 0.03)
bet = index_max(first_order[net].calc_output(last_output[net]))
for i in range(len(first_order[net].stateHiddenNeurons)):
first_order[net].stateHiddenNeurons[i] += randmm(
0.01, 0.03)
wager = index_max(high_order[net].calc_output(
first_order[net].stateHiddenNeurons))
win = deck(bet)
if ((win[0] and wager == 0)):
nbr_gwager += 1
