Skip to main content

Mislabeled samples detection with OP-ELM

Project description

MD-ELM

Detection of originally mislabelled samples in a dataset, with Optimally Pruned Extreme Learning Machine (OP-ELM).

The MDELM function is the core of MD-ELM method, which returns 'likelihood of being a mislabel' score for each sample.

Additional methods are given for running the whole methodology. They generate multiple models, store them in files,
process the models and combine results. Here is an example code to use them:

X,Y = cPickle.load(open("data.pkl","rb"))
mfiles = build_models(X,Y, X.shape[0]/10, k=4, path="./try")

# run all experiments
for data in mfiles:
for elm in data:
run_model(elm)

scores = np.zeros((X.shape[0],))
for data in mfiles:
found = analyze_models(data)
scores[found] += 1
print scores
print "done"

Model files from path="./try" folder can be processed independently with run_model() function on different machines.

Project details


Release history Release notifications

This version
History Node

0.61

History Node

0.42

History Node

0.41

History Node

0.6

History Node

0.5

History Node

0.4

History Node

0.3

History Node

0.2

History Node

0.1

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Filename, size & hash SHA256 hash help File type Python version Upload date
MD-ELM-0.61.tar.gz (9.6 kB) Copy SHA256 hash SHA256 Source None Mar 21, 2014

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging CloudAMQP CloudAMQP RabbitMQ AWS AWS Cloud computing Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page