Unsupervised Clustered Multitask Learning
Scroll ↓
Multitask Clustering
In this research project, I worked on a clustered multitask estimation algorithm. It relies on two types of agents, cluster agents seeking to estimate the cluster centroids as well as the graph structure, and regular agents estimating their respective objective vectors.
The small dots (agents) are estimating collaboratively the location of the diamonds (objective vectors) of the same color.