return cross_entropy
model = VGG19(3)
# model =ResNet50(3)
opt = keras.optimizers.Adam(learning_rate=0.0001)
# opt = keras.optimizers.SGD(learning_rate=1,momentum=0.001)
model.compile(optimizer=opt,
loss="categorical_crossentropy",
metrics='accuracy')
# model.compile(optimizer =opt, loss=weighted_categorical_crossentropy, metrics='accuracy')
model.summary()
dt = DataLoader()
X_Data, Y_Data = dt.LoadTrainingData(
"D:\Sergey\FluorocodeMain\FluorocodeMain\DataForTraining.npz")
x_v, y_v = dt.LoadTrainingData(
"D:\Sergey\FluorocodeMain\FluorocodeMain\DataForValidation.npz")
history = model.fit(X_Data,
Y_Data,
batch_size=4,
shuffle=True,
epochs=100,
validation_data=(x_v, y_v))
plt.plot(history.history['accuracy'])
plt.plot(history.history['val_accuracy'])
plt.title('Model accuracy')
# opt = keras.optimizers.Adam(learning_rate=0.001)
opt = keras.optimizers.SGD(learning_rate=0.003, momentum=0.01)
model.compile(optimizer=opt,
loss="categorical_crossentropy",
metrics='accuracy')
model.summary()
mcp_save = keras.callbacks.ModelCheckpoint(
'mdl_wts_step_fullgenomeStep2-Extra.hdf5',
save_best_only=True,
monitor='val_accuracy',
mode='max')
dt = DataLoader()
# X_Data ,Y_Data = dt.LoadTrainingData("D:\Sergey\FluorocodeMain\FluorocodeMain\DataForTraining1D.npz")
x_v, y_v = dt.LoadTrainingData(
"D:\Sergey\FluorocodeMain\FluorocodeMain\DataForValidation1D.npz")
history = model.fit(x_v,
y_v,
batch_size=8,
epochs=15,
validation_data=(x_v, y_v),
callbacks=[mcp_save, RealDataEval(x_v, y_v)])
plt.plot(history.history['accuracy'])
plt.plot(history.history['val_accuracy'])
plt.title('Model accuracy')
plt.ylabel('Accuracy')
pixelsz = 39.68
ResEnhancement = 2
# genomes = ['Other','CP014051','CP039296','CP000948']
# genomes = ['NC_000913.3','Other','CP014051','CP014787']
genomes = ['NC_000913.3', 'CP014051', 'CP014787', 'NZ_CP009467.1']
# genomes = ['CP014787']
# genomes = ['CP000948']
# genomes = ['CP014051','CP039296', 'CP009467.1','CP014787','CP000948']
AllLabels = []
AllProfiles = []
numsamples = 5000
Dt = DataConverter()
Ds = DataLoader()
for genome in genomes:
EffLabeledTraces = []
SIMTRC = SIMTraces.TSIMTraces(genome, 1.66, 0.34, 0, 'TaqI', pixelsz)
Gauss = Misc.GetGauss1d(
size, Misc.FWHMtoSigma(Misc.GetFWHM(Wavelength, NA, ResEnhancement)),
pixelsz)
if genome != 'Other':
Profiles = []
ReCuts = SIMTRC.GetTraceRestrictions()
ReCutsInPx = SIMTRC.GetDyeLocationsInPixel(ReCuts)
R = RTG.RandomTraceGenerator(Misc.kbToPx(100000, SIMTRC),
Misc.kbToPx(20000, SIMTRC), numsamples)
"""
import Core.Misc as Misc
import Core.SIMTraces as SIMTraces
import Core.RandomTraceGenerator as RTG
import numpy as np
from ImGen.ImGen import TrainImageGenerator
from Core.DataHandler import DataConverter
from Core.DataHandler import DataLoader
genomes = ['CP014787', 'CP000948']
ImagesForAllGenomes = []
LabelsForAllGenomes = []
Dt = DataConverter()
Ds = DataLoader()
for genome in genomes:
SIMTRC = SIMTraces.TSIMTraces(genome, 1.75, 0.34, 0, 'TaqI', 80)
ReCuts = SIMTRC.GetTraceRestrictions()
ReCutsInPx = SIMTRC.GetDyeLocationsInPixel(ReCuts)
ReCutsInPx = ReCutsInPx[1:7000]
R = RTG.RandomTraceGenerator(Misc.kbToPx(40000, SIMTRC),
Misc.kbToPx(5000, SIMTRC), 16 * 4000)
Traces = R.stratsample(np.asarray(ReCutsInPx))
EffLabeledTraces = R.GetEffLabelingRate(Traces, 0.85)
IMGEN = TrainImageGenerator('D:\Sergey\TrainDirectory', 800, 16, 256, 510,
1.4, SIMTRC.PixelSize)