We present a new approach to estimating mixture models based on a new inference principle we have proposed: the latent maximum entropy principle (LME). LME is different both from ...
—Human faces are remarkably similar in global properties, including size, aspect ratio, and location of main features, but can vary considerably in details across individuals, ge...
As e-business grows rapidly, interests in design principles for e-business web sites are increasing. A few studies have suggested design principles with concrete design features, ...
The principle of maximum entropy provides a powerful framework for statistical models of joint, conditional, and marginal distributions. However, there are many important distribu...
Predictive user models often require a phase of effortful supervised training where cases are tagged with labels that represent the status of unobservable variables. We formulate a...