This repository contains the code of "Deep Active Learning with Semidefinite Embedding for Batch Sampling" where two datasets are used in the implementation, e.g. MNIST and ROP datasets. In both folders, .npy files are the distance matrices created using "create_distance_matrix" .py files from the output of the SDE component of the project. Then, using different versions of "training_classifier" .py files, the training process that includes active learning can be done. After obtaining the experiment results in terms of test accuracies, the visualizations can be done via "plot_results" .py files.
ahmetdemirkayaee / deep-active-learning-with-semidefinite-embedding-for-batch-sampling Goto Github PK
View Code? Open in Web Editor NEWDeep Active Learning with Semidefinite Embedding for Batch Sampling