Fig. 2.

Results of applying KLIC to four similar datasets. Left: ARI of KLIC applied to each dataset separately (columns ‘A’, ‘B’, ‘C’ and ‘D’) and to all four datasets together (column ‘A + B+C + D’). The ARI is higher in the last column because KLIC can combine information from all the datasets to find a global clustering. Right: kernel weights associated to each dataset, when applying KLIC to all four datasets together. The algorithm is able to recognize that each dataset contains the same amount of information regarding the global clustering, and therefore assigns on average the same weight to each dataset