Skip to main content

An Interface for Active Learning

Project description

Activizer

Interface for Active Learning

About Active Learning

Active learning is the process by which a learning algorithm can query a user interactively to label data points which are close to the decision boundary formed during classification


The primary objective of this project is to build an Interface for Active learning which simplifies the process of chosing algorithms,query strategies and labels This eliminates the task of writing programs for each task The interface helps annotators of various domains to label data in an interactive manner and also provides features of saving the final model and results


Interface for Active Learning supports Image dataset where the user can upload data in Zip format. It supports 3 classifiers and 7 query strategies.

Classifiers

  • KNN Classifier

  • Random Forest Classifier

  • Decision Tree Classifier

Query Strategies

  • Uncertainty Sampling

  • Random Sampling

  • Entropy Sampling

  • Query By Committee(Uncertainty Sampling)

  • Query By Committee(Vote Entropy Sampling)

  • Query By Committee(Max Disagreement Sampling)

  • Query By Committee(Consensus Entropy Sampling)

This project is implemented with the Active Learning package modAL

How to Run

This project requires python 3.x installed on your machine

Installation

The package can be installed using the command : pip install activizer

Open Python console and run the following commands

  • from activizer import app

  • app.run()


  • Copy the url in the browser

  • Select the Classifier Algorithm, the Query Strategy and give the number of samples you wish to label. Then select the training / testing dataset in Zip format


  • For each iteration an image will be shown and a dropdown to label that image. Below them will be shown a graph with current accuracy.


  • After all the iterations are over, Final accuracy with graph will be shown


  • The user can see the images along with the labels provided by the algorithm selected during training. The trained model can be downloaded in pickle format (.pickle file) and can be used for prediction by clicking "Already have a model" on the Main Page. The user can then upload the pickle file and use the model to classify images

  • The Interface can be used for prediction by uploading the validation dataset in Zip format.


  • The result will be shown in a table consisting of image name and label predicted by the model.


Project details


Download files

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

Files for activizer, version 1.0.7
Filename, size File type Python version Upload date Hashes
Filename, size activizer-1.0.7.tar.gz (306.6 kB) File type Source Python version None Upload date Hashes View

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring DigiCert DigiCert EV certificate Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page