Detecting and Reacting to Data-Drift in Streaming Optimisation Domains - Code and Data

This repository contains the code and data for reproducibility of the paper 'Detecting and Reacting to Data-Drift in Streaming Optimisation Domains'. The following files are included: algo_data.zip : contains algorithm raw performance data; Art_gallery_features_500_samples_full.csv : ELA feature data on the Art Gallery problem; average_loss_accuracy_chunk.csv : results of the encoding switching when drift is detected; drift_detection_zenodo.ipynb : jupyter notebook with the code do to generate the incremental drift scenario and detect drift with NannyML (other scenearios can be generated using this base); plot_features.zip : additional plots not available in the paper; Objective_function.zip : singularity containers and binaries of the code used for the objective function. How to launch the objective functions: singularity run drift_num.sif instance_size nb_cameras nb_constraints x0 y0 x1 y1 ... (x,y being the coordinates of the cameras) singularity run drift_bit.sif instance_size nb_cameras nb_constraints bistring (bistring of size instance_size or nb_constraints with the position of the cameras)