def main(cm_file, perm_file, steps, labels_file):
"""Orchestrate."""
# Load confusion matrix
with open(cm_file) as f:
cm = json.load(f)
cm = np.array(cm)
# Load permutation
if os.path.isfile(perm_file):
print("loaded %s" % perm_file)
with open(perm_file) as data_file:
perm = json.load(data_file)
else:
perm = random.shuffle(list(range(len(cm))))
print("Score without perm: {:0.2f}%".format(
calculate_split_error(cm) * 100))
result = simulated_annealing(cm,
perm,
score=calculate_split_error,
deterministic=True,
steps=steps)
# First recursive step
# split_i = int(len(cm) / 2)
# cm = result['cm'][:split_i, :split_i]
# perm = list(range(split_i))
# result = simulated_annealing(cm, perm,
# score=calculate_split_error,
# deterministic=True,
# steps=steps)
print("Score: {}".format(calculate_split_error(result['cm'])))
print("Perm: {}".format(list(result['perm'])))
# Load labels
if os.path.isfile(labels_file):
with open(labels_file) as f:
symbols = json.load(f)
else:
symbols = read_symbols()
print("Symbols: {}".format([symbols[i] for i in result['perm']]))
plot_cm(result['cm'], zero_diagonal=True)
def main(cm_file, perm_file, steps, labels_file):
"""Orchestrate."""
# Load confusion matrix
with open(cm_file) as f:
cm = json.load(f)
cm = np.array(cm)
# Load permutation
if os.path.isfile(perm_file):
print("loaded %s" % perm_file)
with open(perm_file) as data_file:
perm = json.load(data_file)
else:
perm = random.shuffle(list(range(len(cm))))
print("Score without perm: {:0.2f}%".format(calculate_split_error(cm) * 100))
result = simulated_annealing(cm, perm,
score=calculate_split_error,
deterministic=True,
steps=steps)
# First recursive step
# split_i = int(len(cm) / 2)
# cm = result['cm'][:split_i, :split_i]
# perm = list(range(split_i))
# result = simulated_annealing(cm, perm,
# score=calculate_split_error,
# deterministic=True,
# steps=steps)
print("Score: {}".format(calculate_split_error(result['cm'])))
print("Perm: {}".format(list(result['perm'])))
# Load labels
if os.path.isfile(labels_file):
with open(labels_file, "r") as f:
symbols = json.load(f)
else:
symbols = read_symbols()
print("Symbols: {}".format([symbols[i] for i in result['perm']]))
plot_cm(result['cm'], zero_diagonal=True)
X_train = X_train.astype('float32')
X_val = X_val.astype('float32')
X_test = X_test.astype('float32')
X_train /= 255.0
X_val /= 255.0
X_test /= 255.0
# Calculate confusion matrix
y_val_i = y_test.flatten()
y_val_pred = model.predict(X_test)
y_val_pred_i = y_val_pred.argmax(1)
cm = np.zeros((n_classes, n_classes), dtype=np.int)
for i, j in zip(y_val_i, y_val_pred_i):
cm[i][j] += 1
acc = get_accuracy(cm)
print("Accuracy: {:0.2f}".format(acc * 100))
# Create plot
plot_cm(cm)
# Serialize confusion matrix
with open('cm.json', 'w', encoding='utf8') as outfile:
str_ = json.dumps(cm.tolist(),
indent=4,
sort_keys=True,
separators=(',', ':'),
ensure_ascii=False)
outfile.write(to_unicode(str_))
continue
return grouping
with open('confusion-matrix.json') as f:
cm = json.load(f)
cm = np.array(cm)
# Visulize
print("Score: {}".format(calculate_score(cm)))
# Load permutation
perm_file = 'permutations/hasy-58626.json'
if os.path.isfile(perm_file):
with open(perm_file) as data_file:
perm = json.load(data_file)
else:
perm = list(range(len(cm)))
result = simulated_annealing(cm, perm, deterministic=True, steps=1)
print("Score: {}".format(calculate_score(result['cm'])))
print("Accuracy: {}".format(get_accuracy(result['cm'])))
print("Perm: {}".format(list(result['perm'])))
symbols = read_symbols()
grouping = get_group(result['cm'])
for group in grouping:
if len(group) > 1:
print("Symbols: {}".format([symbols[i] for i in group]))
plot_cm(result['cm'], zero_diagonal=True)
continue
return grouping
with open('confusion-matrix.json') as f:
cm = json.load(f)
cm = np.array(cm)
# Visulize
print("Score: {}".format(calculate_score(cm)))
# Load permutation
perm_file = 'permutations/hasy-58626.json'
if os.path.isfile(perm_file):
with open(perm_file) as data_file:
perm = json.load(data_file)
else:
perm = list(range(len(cm)))
result = simulated_annealing(cm, perm, deterministic=True, steps=1)
print("Score: {}".format(calculate_score(result['cm'])))
print("Accuracy: {}".format(get_accuracy(result['cm'])))
print("Perm: {}".format(list(result['perm'])))
symbols = read_symbols()
grouping = get_group(result['cm'])
for group in grouping:
if len(group) > 1:
print("Symbols: {}".format([symbols[i] for i in group]))
plot_cm(result['cm'], zero_diagonal=True)