With more devices connected, sensor data logged and people active in social networks, the trend towardsworking with dynamic data is clear. The number of applications where it becomes essential to perform real timeanalysis on data streams grows accordingly, each with its own challenges. From this area of data stream analysiswe benchmark the performance of current state of the art clustering algorithms: CluStream, DenStream andClusTree. We also adapt a Variational Autoencoder to perform in the context of non-stationary data streamsand assess its generative capabilities for dimensionality reduction. From this limited lab experiment we showthat while there is a significant improvement in the clustering accuracy of high dimensional datasets after adimensionality reduction with a Variational Autoencoder, not all clustering algorithms benefit in the sameway from it. Additionally we show that regardless of the clustering algorithm, no relevant improvement in thepurity of the clusters could be obtained after the dimensionality reduction.
展开▼