KJL

A python library is created for efficient anomaly detection, which mainly includes two submodules: kernel projection ('projection') and novelty detection models ('models').

Architecture

• docs/: includes all documents (such as APIs)

• applications/: includes applications and toy examples and datasets for you to play with it

• kjl/: source codes: includes two main sublibraries (projects and models)

  • projects/: includes KJL and Nystrom (such as OCSVM)
  • models/: includes OCSVM and GMM
  • utils/: includes common functions (such as load and dump data)
  • visul/: includes visualization functions

• thrid_party/: others (such as xxx.sh, make)

• tools/: includes useful tools (e.g., upload files)

• LICENSE.txt

• README.md

• requirements.txt

• setup.py

• version.txt

Install

pip3 install . 
(pip3 will call setup.py to install the library automatically)

Usage

""" 1.1 Parse data and extract features

"""
lg.info(f'\n--- 1.1 Load data')
feat_file = f'{OUT_DIR}/DEMO_IAT+SIZE.dat'
X, y = load(feat_file)

""" 1.2 Split train and test set

"""
lg.info(f'\n--- 1.2 Split train and test set')
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3,
                                                    shuffle=True, random_state=RANDOM_STATE)
lg.debug(f'X_train:{X_train.shape}, y_train: {Counter(y_train)}')
lg.debug(f'X_test:{X_test.shape}, y_test: {Counter(y_test)}')

""" 1.3 preprocessing
	projection
"""
lg.info(f'\n--- 1.3 Preprocessing')
proj = Projection(name='KJL')
proj.fit(X_train)
X_train = proj.transform(X_train)

""" 2.1 Build the model

"""
lg.info(f'\n--- 2.1 Build the model')
model_name = 'OCSVM'
model = Model(name=model_name, q=proj.q, overwrite=OVERWRITE, random_state=RANDOM_STATE)
model.fit(X_train, y_train)

""" 2.2 Evaluate the model

"""
lg.info(f'\n--- 2.2 Evaluate the model')
X_test = proj.transform(X_test)
res = model.eval(X_test, y_test)

""" 3. Dump the result to disk

"""
lg.info(f'\n--- 3. Save the result')
res_file = os.path.join(OUT_DIR, f'DEMO-KJL-{model_name}-results.dat')
check_path(os.path.dirname(res_file))
dump(res, out_file=res_file)
lg.info(f'res_file: {res_file}')

return res

For more examples, please check the 'examples' directory

TODO

• Complete the online application
• Further evaluate and optimize the library continually.
• Add 'test' cases
• Add LICENSE.txt
• Generated docs from docs-string automatically

Welcome to make any comments to make it more robust and easier to use!

Contact

• Email: [email protected]