Beyond I-Con: Exploring New Dimension of Distance Measures in Representation Learning