Repository for the batch-incremental UMAP-kNN algorithm implemented in Scikit-multiflow.
For more informations about Scikit-multiflow, check out the official website:
https://scikit-multiflow.github.io/
To cite the UMAP-kNN in a publication, please cite the following paper:
Maroua Bahri, Bernhard Pfahringer, Albert Bifet, Silviu Maniu. Efficient Batch-Incremental Classification for Evolving Data Streams. In the Symposium on Intelligent Data Analysis (IDA), 2020.
The implementation used in this work is the following:
- batchIncrementalUMAP.py: an example of the batch-incremental UMAP-kNN application.
If you wish to test the UMAP-kNN, you can update the following parameters:
- k: the number of neigbors for kNN
- batch: the batch size
- d: the output dimensionality
- w: the maximum number of instances to store inside the sliding window
- stream: the data stream
The datasets used in this work are compressed and available at the root directory.