Online Partitioned Local Depth for semi-supervised applications
Researchers have developed an extension of the partitioned local depth (PaLD) algorithm, named online PaLD, designed for semi-supervised applications. This new approach allows for the efficient extension of a pre-computed cohesion network to new data points, making larger datasets accessible for exact analysis. Initial applications demonstrate its potential in online anomaly detection and semi-supervised classification within healthcare datasets. AI
IMPACT Introduces a more scalable method for analyzing larger datasets in semi-supervised learning tasks.