ada/nn420-private-pine64backup
Archived
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Fix this indention

Browse source
This commit is contained in:
Alexander Huddleston 2017-12-07 22:08:36 -06:00
parent fef852dbc7
commit ae19d80d7d
1 changed files with 47 additions and 43 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

90
neural_network.py
View file

@ -81,7 +81,7 @@ def format_man_dist(elem):
target = [] target = []
for i in range(29): for i in range(29):
filename = join('data', str(i) + '.bin') filename = join('/pub/faculty_share/daugher/datafiles/data/' + str(i) + 'states.bin')
# Debugging to print the current file from which states are being parsed. # Debugging to print the current file from which states are being parsed.
#print(i) #print(i)
@ -91,7 +91,7 @@ for i in range(29):
data = f.read(8) data = f.read(8)
counter = 0 counter = 0
while(data and counter < 1000): while(data and counter < 2000):
temp.append(format_man_dist(i)) temp.append(format_man_dist(i))
data = f.read(8) data = f.read(8)
@ -123,7 +123,7 @@ model.compile(optimizer='sgd',
for i in range(29): for i in range(29):
filename = join('data', str(i) + '.bin') filename = join('/pub/faculty_share/daugher/datafiles/data/' + str(i) + 'states.bin')
# Debugging to print the current file from which states are being parsed. # Debugging to print the current file from which states are being parsed.
print(i) print(i)
@ -133,7 +133,7 @@ for i in range(29):
counter = 0 counter = 0
training = [] training = []
while(data and counter < 1000): while(data and counter < 2000):
bin_data = reduce(format_input, list(data), []) bin_data = reduce(format_input, list(data), [])
bin_data.reverse() bin_data.reverse()
bin_data = bin_data[16:] bin_data = bin_data[16:]
@ -146,70 +146,74 @@ for i in range(29):
#print(training[0]) #print(training[0])
# Train the network. # Train the network.
model.fit(np.array(training), np.array(target[i]), epochs=5, batch_size=1000) model.fit(np.array(training), np.array(target[i]), epochs=8, batch_size=2000)
#model.train_on_batch(np.array(temp), np.array(target)) #model.train_on_batch(np.array(temp), np.array(target))
# Used for testing data # Used for testing data
with open('data/18.bin', 'rb') as f: for i in range(11, 29):
filename = join('/pub/faculty_share/daugher/datafiles/data/', str(i) + '.bin')
for i in range(1000): print(i)
data = f.read(8)
data = f.read(8) with open('/pub/faculty_share/daugher/datafiles/data/18states.bin', 'rb') as f:
counter = 0 for i in range(2000):
data = f.read(8)
testing = [] data = f.read(8)
testing_target = [] counter = 0
while(data): testing = []
bin_data = reduce(format_input, list(data), [])
bin_data.reverse()
bin_data = bin_data[16:]
testing.append(bin_data) testing_target = []
pos_data = reduce(format_pos, enumerate(list(data)), []) while(data):
pos_data.reverse() bin_data = reduce(format_input, list(data), [])
pos_data = pos_data[1:] bin_data.reverse()
bin_data = bin_data[16:]
state_pos = [] testing.append(bin_data)
for p in pos_data: pos_data = reduce(format_pos, enumerate(list(data)), [])
state_pos.append(p[1]) pos_data.reverse()
pos_data = pos_data[1:]
testing_target_pos = reduce(generate_pos, pos_data, []) state_pos = []
testing_target.append(format_man_dist(man_dist_state(state_pos, testing_target_pos))) for p in pos_data:
state_pos.append(p[1])
counter += 1 testing_target_pos = reduce(generate_pos, pos_data, [])
data = f.read(8)
#print(testing_target) testing_target.append(format_man_dist(man_dist_state(state_pos, testing_target_pos)))
# Evaluate accuracy counter += 1
data = f.read(8)
loss_and_metrics = model.evaluate(np.array(testing),np.array(testing_target), batch_size=1000)
# Generating predictions: # Evaluate accuracy
predictions = model.predict(np.array(testing), batch_size=1000) loss_and_metrics = model.evaluate(np.array(testing),np.array(testing_target), batch_size=1000)
output = [] # Generating predictions:
for p in range(len(predictions)): predictions = model.predict(np.array(testing), batch_size=1000)
if np.argmax(testing_target[p]) < 18:
output.append(100*((18 - (28 - np.argmax(predictions[p]))) / (18 - np.argmax(testing_target[p]))))
else:
output.append(0)
#for i in range(len(output)): output = []
# print(output[i])
print(np.array(output).mean()) for p in range(len(predictions)):
if np.argmax(testing_target[p]) < 18:
output.append(100*((18 - (28 - np.argmax(predictions[p]))) / (18 - np.argmax(testing_target[p]))))
else:
output.append(0)
print(loss_and_metrics) #for i in range(len(output)):
# print(output[i])
print(model.metrics_names) print("Percentage possible improvement: ", np.array(output).mean())
print(model.metrics_names[0], loss_and_metrics[0])
print(model.metrics_names[1], loss_and_metrics[1])