def __init__(self):
print "ann_io constructor called"
self.__data = ANN()
def __init__(self):
print "ann_io constructor called"
self.__data = ANN()
class ann_io:
def read(self):
file = open("ann.txt")
data = file.read()
__exec_order = False
__new_layer = False
__new_link = False
print "ann_io: Begin parsing script"
for strdat in data.split('\n'):
str = strdat.split()
if not str:
continue
if str[0] == "begin" and str[1] == "exec_order":
__exec_order = True
continue
elif str[0] == "begin" and str[1] == "layer":
self.__layer = ANN_LAYER()
__new_layer = True
continue
elif str[0] == "begin" and str[1] == "link":
self.__link = ANN_LINK()
__new_link = True
continue
elif str[0] == "end" and str[1] == "exec_order":
__exec_order = False
continue
elif str[0] == "end" and str[1] == "layer":
self.__data.add_layer(self.__layer)
__new_layer = False
continue
elif str[0] == "end" and str[1] == "link":
self.__data.add_link(self.__link)
__new_link = False
continue
if not __exec_order:
self.__parse(str)
else:
print str[0]
self.__parse_exec_order(str[0])
return self.__data
def __init__(self):
print "ann_io constructor called"
self.__data = ANN()
def __parse_exec_order(self,s):
self.__data.add_exec_order(s)
def __parse(self,s):
if s[0] == "layer_name":
self.__layer.set_layer_name(s[1])
elif s[0] == "layer_type":
if s[1] == "input":
self.__layer.set_layer_type(TYPE.INPUT)
elif s[1] == "hidden":
self.__layer.set_layer_type(TYPE.HIDDEN)
elif s[1] == "output":
self.__layer.set_layer_type(TYPE.OUTPUT)
elif s[0] == "layer_size":
self.__layer.set_layer_size(int(s[1]))
elif s[0] == "layer_act_func":
if s[1] == "sigmoid":
self.__layer.set_layer_act_func(Layer.sigmoid)
elif s[1] == "dsigmoid":
self.__layer.set_layer_act_func(Layer.dsigmoid)
elif s[0] == "link_name_pre":
self.__link.set_link_name_pre(s[1])
elif s[0] == "link_name_post":
self.__link.set_link_name_post(s[1])
elif s[0] == "link_conn_top":
if s[1] == "full":
self.__link.set_link_conn_top(TOPOLOGY.FULL)
elif (s[0] == "oneone"):
self.__link.set_link_conn_top(TOPOLOGY.ONEONE)
elif s[0] == "triangular":
self.__link.set_link_conn_top(TOPOLOGY.TRIANGULAR)
elif s[0] == "stochastic":
self.__link.set_link_conn_top(TOPOLOGY.STOCHASTIC)
elif s[0] == "link_conn_prob":
self.__link.set_link_conn_prob(float(s[1]))
elif s[0] == "link_learn_rule":
if s[1] == "general-hebb":
self.__link.set_link_learn_rule(RULE.GENERAL)
elif s[1] == "oja":
self.__link.set_link_learn_rule(RULE.OJA)
elif s[0] == "link_learn_param":
self.__link.set_link_learn_param(float(s[1]))
elif s[0] == "link_learn_rate":
self.__link.set_link_learn_rate(float(s[1]))
elif s[0] == "link_range":
t = s[1].split(",")
(min,max) = (t[0],t[1])
self.__link.set_link_range((min,max))
class ann_io:
def read(self):
file = open("ann.txt")
data = file.read()
__exec_order = False
__new_layer = False
__new_link = False
__new_arc = False
print "ann_io: Begin parsing script"
for strdat in data.split('\n'):
str = strdat.split()
if not str:
continue
if str[0] == "begin" and str[1] == "exec_order":
__exec_order = True
continue
elif str[0] == "begin" and str[1] == "layer":
self.__layer = ANN_LAYER()
__new_layer = True
continue
elif str[0] == "begin" and str[1] == "link":
self.__link = ANN_LINK()
__new_link = True
continue
elif str[0] == "begin" and str[1] == "arc":
__new_arc = True
self.__data.set_hard_wired(True)
continue
elif str[0] == "end" and str[1] == "exec_order":
__exec_order = False
continue
elif str[0] == "end" and str[1] == "layer":
self.__data.add_layer(self.__layer)
__new_layer = False
continue
elif str[0] == "end" and str[1] == "link":
self.__data.add_link(self.__link)
__new_link = False
continue
elif str[0] == "end" and str[1] == "arc":
__new_arc = False
if not __exec_order:
self.__parse(str)
else:
print str[0]
self.__parse_exec_order(str[0])
return self.__data
def __init__(self):
print "ann_io constructor called"
self.__data = ANN()
def __parse_exec_order(self, s):
self.__data.add_exec_order(s)
def __parse(self, s):
if s[0] == "layer_name":
self.__layer.set_layer_name(s[1])
elif s[0] == "layer_type":
if s[1] == "input":
self.__layer.set_layer_type(TYPE.INPUT)
elif s[1] == "hidden":
self.__layer.set_layer_type(TYPE.HIDDEN)
elif s[1] == "output":
self.__layer.set_layer_type(TYPE.OUTPUT)
elif s[0] == "layer_size":
self.__layer.set_layer_size(int(s[1]))
elif s[0] == "layer_act_func":
if s[1] == "sigmoid":
self.__layer.set_layer_act_func(Layer.sigmoid)
elif s[1] == "dsigmoid":
self.__layer.set_layer_act_func(Layer.dsigmoid)
elif s[1] == "right_sigmoid":
self.__layer.set_layer_act_func(Layer.right_sigmoid)
elif s[1] == "linear":
self.__layer.set_layer_act_func(Layer.linear)
elif s[1] == "plinear":
self.__layer.set_layer_act_func(Layer.plinear)
elif s[1] == "step":
self.__layer.set_layer_act_func(Layer.step)
elif s[1] == "logistical":
self.__layer.set_layer_act_func(Layer.logistical)
elif s[0] == "link_name_pre":
self.__link.set_link_name_pre(s[1])
elif s[0] == "link_name_post":
self.__link.set_link_name_post(s[1])
elif s[0] == "link_conn_top":
if s[1] == "full":
self.__link.set_link_conn_top(TOPOLOGY.FULL)
elif (s[1] == "oneone"):
self.__link.set_link_conn_top(TOPOLOGY.ONEONE)
elif s[1] == "triangular":
self.__link.set_link_conn_top(TOPOLOGY.TRIANGULAR)
elif s[1] == "stochastic":
self.__link.set_link_conn_top(TOPOLOGY.STOCHASTIC)
elif s[0] == "link_conn_prob":
self.__link.set_link_conn_prob(float(s[1]))
elif s[0] == "link_learn_rule":
if s[1] == "general-hebb":
self.__link.set_link_learn_rule(RULE.GENERAL)
elif s[1] == "classical-hebb":
self.__link.set_link_learn_rule(RULE.CLASSICAL)
elif s[1] == "oja":
self.__link.set_link_learn_rule(RULE.OJA)
elif s[1] == "none":
self.__link.set_link_learn_rule(RULE.NONE)
elif s[0] == "link_learn_param":
self.__link.set_link_learn_param(float(s[1]))
elif s[0] == "link_learn_rate":
self.__link.set_link_learn_rate(float(s[1]))
elif s[0] == "link_range":
t = s[1].split(",")
(min, max) = (t[0], t[1])
elif s[0] == "arc":
t = s[1].split(",")
from_node, to_node, weight = int(t[0]), int(t[1]), float(t[2])
self.__link.add_arc(ANN_ARC(from_node, to_node, weight))