def init(self, ):
print('SOM Server init')
self.host = 'localhost'
self.port = 7000
self.som = som.SOM()
self.som.open('cfg/test.json')
def input_to_tsp(self):
os.system('cls' if os.name == 'nt' else 'clear')
print(self.headerText)
print("--- TESTING TCP SCENARIO ---")
problemNumber = input('Choose which problem (number): ')
epochs = input("Epochs: ")
eta = input("Initial learning rate: ")
tauEta = input("Learning rate time constant: ")
sigma = input("Initial neighbourhood size: ")
tauSigma = input("Neighbourhood time constant: ")
plotInterval = input("Plot interval: ")
time.sleep(0.7)
confirmInput = input("Proceed with the chosen parameters[y/n]? ")
if confirmInput == 'y':
tspSOM = som.SOM(problemType='TSP',
problemArg=int(problemNumber),
epochs=int(epochs),
sigma_0=float(sigma),
tau_sigma=float(tauSigma),
eta_0=float(eta),
tau_eta=float(tauEta),
plotInterval=int(plotInterval))
tspSOM.run()
wait = input("PRESS ENTER TO EXIT TO MAIN MENU")
self.state = "options"
def config_to_tsp(self):
os.system('cls' if os.name == 'nt' else 'clear')
print(self.headerText)
print("--- LOAD BEST PARAMETER SCENARIO FOR TSP --- ")
print("Supported config formats: .txt\n")
print("Available scenarios:")
for root, dirs, files in os.walk("../configs/tsp"):
for file in files:
if file.endswith('.txt'):
print(file)
fileName = input("\nSelect data source: ")
confirmInput = input("Proceed with the chosen parameters[y/n]? ")
if confirmInput == 'y':
with open('../configs/tsp/' + fileName, 'r') as f:
for paramLine in f:
paramLine = paramLine.strip("\n")
paramLine = paramLine.split(",")
if (paramLine[0] == ''):
continue # Skip empty lines
elif (paramLine[0][0] == '#'):
continue # Skip comments
else:
paramName = paramLine[0]
paramLine.pop(0)
if paramName == 'problemNumber':
problemNumber = int(paramLine[0])
elif paramName == 'plotInterval':
plotInterval = int(paramLine[0])
elif paramName == 'epochs':
epochs = int(paramLine[0])
elif paramName == 'sigma_0':
sigma_0 = float(paramLine[0])
elif paramName == 'tau_sigma':
tau_sigma = int(paramLine[0])
elif paramName == 'eta_0':
eta_0 = float(paramLine[0])
elif paramName == 'tau_eta':
tau_eta = int(paramLine[0])
else:
raise AssertionError(
"Parameter: " + paramName +
", is not a valid parameter name.")
tspSOM = som.SOM(problemType='TSP',
problemArg=problemNumber,
plotInterval=plotInterval,
epochs=epochs,
sigma_0=sigma_0,
tau_sigma=tau_sigma,
eta_0=eta_0,
tau_eta=tau_eta)
tspSOM.run()
wait = input("PRESS ENTER TO EXIT TO MAIN MENU")
self.state = "options"
pass
def build_SOM(db_path, query, N, path = None, random_seed=21893698, dimensions = None, \
normalization_mode = 'None', ID = False, initial_weights = None,
lr = 1, sigma = 0.5,
cyclic = None, cyclic_norms = None, cyclic_modes = None):
# Create a som
with get_data.EDataDB(db_path) as db:
db.open()
db.c.execute(query)
data = db.c.fetchall()
print('Analysing %d points...' % (len(data), ))
somap = som.SOM(data,
random_seed=random_seed,
dimensions=dimensions,
normalization_mode=normalization_mode,
ID=ID,
learning_rate=lr,
sigma=sigma,
cyclic=cyclic,
cyclic_norms=cyclic_norms,
cyclic_modes=cyclic_modes)
# Initialize weights
if initial_weights is None:
somap.random_weights_init(somap.data)
if initial_weights is not None:
somap._weights = initial_weights
# Train SOM
somap.train(N)
# Generate informative maps
somap.generate_distance_map()
somap.generate_activation_response()
# # Build winners and members
# somap.build_winners()
# somap.build_cluster_members()
# Save SOM
if path != None:
with open(path, 'wb') as out_f:
pickle.dump(somap, out_f)
return somap
def input_to_icp(self):
os.system('cls' if os.name == 'nt' else 'clear')
print(self.headerText)
print("--- TESTING ICP SCENARIO ---")
epochs = input("Epochs: ")
gridSize = input("Grid size: ")
eta = input("Initial learning rate: ")
tauEta = input("Learning rate time constant: ")
sigma = input("Initial neighbourhood size: ")
tauSigma = input("Neighbourhood time constant: ")
plotInterval = input("Plot interval: ")
testInterval = input("Test interval: ")
nmbrOfCases = input("Number of training cases to use: ")
fillIn = input("Fill in non-classified nodes wrt neighbours[y/n]: ")
if fillIn == "y":
fillIn = True
else:
fillIn = False
time.sleep(0.7)
confirmInput = input("Proceed with the chosen parameters[y/n]? ")
if confirmInput == 'y':
icpSOM = som.SOM(problemType='ICP',
problemArg=None,
gridSize=int(gridSize),
initialWeightRange=(0, 1),
epochs=int(epochs),
sigma_0=float(sigma),
tau_sigma=float(tauSigma),
eta_0=float(eta),
tau_eta=float(tauEta),
plotInterval=int(plotInterval),
testInterval=int(testInterval),
fillIn=True,
nmbrOfCases=int(nmbrOfCases))
icpSOM.run()
wait = input("PRESS ENTER TO EXIT TO MAIN MENU")
self.state = "options"
training_set_df = pd.read_csv(training_set_path)
print(
time.strftime("%H:%M:%S") + ' training set shape: ' +
str(training_set_df.shape))
# find all the ohlc files as well
ohlc_files = ohlc_file_helper.get_files_in_directory(ohlc_files_path, '.csv')
ohlc_date_map = list(ohlc_file_helper.build_ohlc_date_map(ohlc_files))
print(time.strftime("%H:%M:%S") + ' training SOM')
training_set = training_set_df.iloc[:100, 1:(example_len + 1)].values
print(training_set)
num_training = 20 #400
sm = som.SOM(20, 30, example_len, num_training)
sm.train(training_set)
# plt.imshow(np.reshape(s.get_weights(), [30, 30, 3]))
# plt.show()
#Get output grid
image_grid = sm.get_centroids()
#Plot
plt.imshow(image_grid)
plt.title('SOM')
plt.show()
def setUp(self):
self.som = som.SOM([])
self.som.col_sz = 4
self.som.row_sz = 4
extracted = pd.DataFrame()
for label in col_label:
extracted[label] = foods[label]
features = extracted.iloc[:, 3:].as_matrix()
vs = []
for i in range(len(extracted.index)):
vs.append(som.NamedVector(extracted.iloc[i, 1], extracted.iloc[i, 3:].as_matrix()))
random.shuffle(vs)
for v in vs:
v.dump()
msom = som.SOM(vs)
#msom.arranged()
#msom.fission()
msom.execute_with_animation('result.txt')
#-----------
#while msom.update_som_assign():
# pass
#msom.reassign_all()
#-----------
#-----------
#msom.update_som_repeatedly(200)
#msom.clustering()
#-----------
# TODO: Spiking SOM from pyqtgraph.Qt import QtGui, QtCore import som import visualization import dataset # Obtain data from video and visualize it data_set_instance = dataset.DataSet() data_set_instance.visualize_data() data = data_set_instance.merge_data_set() # Perform SOM training som_map = som.SOM(data, proto_dim=2, learning_rate_init=0.001, save_after=None) visual = visualization.SomVisualization(som=som_map) visual.run()