Abstract. We introduce a Forward Backward and Model Selection algorithm (FBMS) for constructing a hybrid regression network of radial and perceptron hidden units. The algorithm det...
In this work, we take a novel line of approaches to evolve images. It is motivated by the total variation method, known for its denoising and edge-preserving effect. Our approach ...
Gait recognition is an effective approach for human identification at a distance. During the last decade, the theory of hidden Markov models (HMMs) has been used successfully in th...
The main problem in any model-building situation is to choose from a large set of covariates those that should be included in the "best" model. A decision to keep a vari...
Zoran Bursac, C. Heath Gauss, David Keith Williams...
This paper presents a method for estimating geographic
location for sequences of time-stamped photographs. A
prior distribution over travel describes the likelihood of
traveling...
Evangelos Kalogerakis, Olga Vesselova, James Hays,...