max k [K] hik(x)>1 B/K1 min i [n ] min l [K] aill(x)/K. Step2. Weassumethathjr(x)attainsthelargestvalueofhik(x)foranyi [n],k [K]. Then hjr(x)>1 min

A.1 ImplementationDetails Network Architecture: Inspired by [33], we utilize a pre-trained ResNet-50 [20] as the feature extractor for object recognition tasks (i.e., Office-31 [22], Office-Caltech [18] and Office-Home [46]). Theoverallframeworkis trained under an end-to-end manner via back-propagation. The stochastic gradient descent with momentum value as 0.9 is employed as the network optimizer. The initial learning rates for feature extractor and bottleneck layer are respectively set as 10 3 and 10 2, while the parameters of classifier are frozen. It is exponentially decayed as the training process.