Connecting Jensen–Shannon and Kullback–Leibler Divergences: A New Bound for Representation Learning | Read Paper on Bytez