All Dates/Times are Australian Eastern Standard Time (AEST)

Technical Program

Paper Detail

Paper IDD5-S7-T3.3
Paper Title Synthesizing New Expertise via Collaboration
Authors Bijan Mazaheri, Siddharth Jain, Jehoshua Bruck, California Institute of Technology, United States
Session D5-S7-T3: Classification I
Chaired Session: Saturday, 17 July, 00:00 - 00:20
Engagement Session: Saturday, 17 July, 00:20 - 00:40
Abstract Consider a set of classes and an uncertain input. Suppose, we do not have access to data and only have knowledge of perfect pairwise classifiers between a few classes in the set. How can we use this to gain knowledge about unknown pairwise classifiers? In this paper, we define a framework to analyze this problem. In particular, we define a knowledge graph where classes denote vertices and edges that are present only between classes with known classifiers. We derive necessary conditions on the edge weights for a knowledge graph to be valid. Further, we show that these conditions are also sufficient if the graph is a cycle, which can yield unintuitive results even when given perfect classifiers. We provide an algorithm to obtain upper and lower bounds on the weights of unknown edges in a knowledge graph.