PURPOSE The global prevalence of age-related macular degeneration and associated central vision loss is rising. Central vision loss hinders the performance of many activities of daily living. Adaptive strategies such as eccentric viewing and steady eye strategy may be used to compensate for central vision loss. In order to establish the potential of these rehabilitation strategies, this systema...

Abstract Spatial transform networks (STN) are widely used since they can images captured from different viewpoints to obtain an objective image. These use image any viewpoint as input and the desired a label. Usually, these segmented, but this could lead convergence problems if percentage of overlap between segmented is quite low. In paper, we propose new training method facilitate STN in cases...

Imputing missing values in high dimensional time-series is a difficult problem. This paper presents a strategy for training energy-based graphical models for imputation directly, bypassing difficulties probabilistic approaches would face. The training strategy is inspired by recent work on optimization-based learning (Domke, 2012) and allows complex neural models with convolutional and recurren...