async3secSegments=Segmentation("None",periodSync=False,sizeType="fixed",frameSizeMs=3000.0,hopSizeMs=1000.0)
segStrategies=[async2secSegments,async3secSegments]
#Define features to be used
features=[]
for featName in ['SubEnv']:#other options: 'MFCC','MelSpec'
for segType in segStrategies:
for timeDim in [32,64]:
for freqDim in [16]:
features.append(Feature(featName,[timeDim,freqDim],"frame",segType,involveDelta=False))
#Define data specifications for this database
data=Data(dbaName,dataFolder,featureFolder,features,useBalancedData,splitRatios,info)
#Defining NN model with a name.
# Implementation is in models.py. Feel free to add your own models and
# test by just changing the name here
modelNames=['uocSeq1','uocSeq2']
#Running random split and testing several times (1/testSetPercentage)
# ex: if test set is 20%, tests will be repeated 5 times
numExperiments=int(1/splitRatios[-1])
for i in range(numExperiments):
for modelName in modelNames:
#Define test specifications and run
singleTest=Test(modelName,data,resultsFolder,batch_size=128,num_epochs=50)
#Run the tests: outputs will be put in the results folder
singleTest.run()
#Cleaning this test sessions' intermediate files
cleanFilesOfPrevSess([dataFolder])
def test_agent(args, shared_value, share_net):
test = Test(args, shared_value, share_net)
test.run()
from torch.optim import lr_scheduler
import numpy as np
import torchvision
from torchvision import datasets, models, transforms
import matplotlib.pyplot as plt
import time
import os
import copy
from utils import get_device
from dataset.DatasetHelper import DatasetHelper
from Test import Test
if __name__ == '__main__':
train = Training()
train.train(n_epochs=5,
batch_size=64,
model_type=ModelType.cnn,
criterion=nn.CrossEntropyLoss(),
learning_rate=0.001)
train.train(n_epochs=5,
batch_size=64,
model_type=ModelType.fine_tuned,
criterion=nn.BCELoss(),
learning_rate=0.001,
enable_scheduler=True)
test = Test()
test.run()
def test_agent(args, shared_queue,shared_value):
test = Test(args, shared_queue,shared_value)
test.run()
def takeTest():
chapter = input("what chapter would you like to test on")
test = Test(chapter)
score = test.run()
student.saveScore(chapter, score)
def main(fn):
# data= pd.read_csv("../data/kddcup.data_10_percent_corrected", names=cols)
data = pd.read_csv(fn, header=-1)
# data= remove_missing(data)
# data= impute_missing(data)
data = impute_missing2(data)
# Features to be used in classification
features = [x for x in range(1, len(data.columns))]
X = data[features]
y = data[0]
#h= TGaussianNB(X, y)
#h.run()
print("GaussianNB")
h = Test(X, y, GaussianNB())
h.run()
h.report(fn="../Report/results/cancer.gnb.cm.tex")
s = Search(X, y, GaussianNB(), [{}])
s.search()
s.report("../Report/results/cancer.gnb.tex")
print("DTree neu")
parameters = [{
'criterion': ['gini', 'entropy'],
'max_features': ['auto', 'sqrt', 'log2']
}]
s = Search(X, y, DTree(), parameters)
s.search()
s.report("../Report/results/cancer.dt.tex")
h = Test(X, y,
DTree(max_features='log2', criterion='gini', random_state=1234))
h.run()
h.report(fn="../Report/results/cancer.dt.cm.tex")
print("RF")
parameters = [{
'n_estimators': range(1, 15),
'criterion': ['gini', 'entropy'],
'max_features': ['auto', 'sqrt', 'log2']
}]
s = Search(X, y, RandomForestClassifier(), parameters)
s.search()
s.report("../Report/results/cancer.rf.tex", )
h = Test(
X, y,
RandomForestClassifier(n_estimators=6,
criterion='gini',
max_features='sqrt',
random_state=1234))
h.run()
h.report(fn="../Report/results/cancer.rf.cm.tex")
parameters = [{
'kernel': ['linear', 'sigmoid', 'rbf', 'poly'],
'C': [0.1, 1, 10, 11, 20]
}]
print("SVM")
from sklearn import preprocessing
X_scaled = preprocessing.scale(X)
s = Search(X_scaled, y, SVC(), parameters)
s.search()
s.report("../Report/results/cancer.svm.tex")
h = Test(X, y, SVC(C=11, kernel='poly'))
h.run()
h.report(fn="../Report/results/cancer.svm.cm.tex")
print("KNeighborsClassifier")
parameters = [{
'n_neighbors': range(4, 8),
'weights': ['uniform', 'distance'],
'p': [1, 2]
}]
s = Search(X, y, KNeighborsClassifier(), parameters)
s.search()
s.report("../Report/results/cancer.knn.tex")
h = Test(X, y, KNeighborsClassifier(n_neighbors=5, weights='uniform', p=2))
h.run()
h.report(fn="../Report/results/cancer.knn.cm.tex")
