Foundations of Vision

Abstract

This article discusses the integral nature of theology and spirituality in the writings and practices of John and Charles Wesley. It describes works of piety and works of mercy as a holistic foundation upon which the Wesleys built their movement of renewal in the Church of England in the eighteenth century. Particular attention is given to the means of grace -prayer, biblical engagement, Christian fellowship, and Eucharist, as well as the connection between the sacrament and mission. This article was the first of two lectures originally delivered at Booth College (Winnipeg) as part of the inaugural Earl Robinson Lectures series.

Convolutional neural network models applied to neuronal responses in macaque V1 reveal limited nonlinear processing

“…Work from our own research group has shown that binocular rivalry competition, spatial and feature-based attention, object-based attentional selection, and visual working memory all lead to powerful modulatory effects in the human primary visual cortex ( Cohen & Tong, 2015 ; Harrison & Tong, 2009 ; Jehee, Brady, & Tong, 2011 ; Kamitani & Tong, 2005 ; Tong & Engel, 2001 ). By contrast, a limitation of feedforward neural network models is their inability to account for top–down effects of attention and other task-based goals ( Kay, Bonnen, Denison, Arcaro, & Barack, 2023 ; Tong, 2018 ). It will be of considerable interest for future studies to explore whether variations in CNN architecture, the incorporation of recurrent or top–down processing, or the expansion of stimuli and methods used for network training can further improve the ability of CNN models to predict the nonlinear response properties of V1.…”

An autonomous framework for interpretation of 3D objects geometric data using 2D images for application in additive manufacturing

“…The next step is to categorize these simpler shapes regarding their similarity with familiar cases. To do so, CNNs are the best candidate to use, as they can do the recognition process with a surprisingly high accuracy ( Tong, 2018 ). Although there are 3D CNNs which could be used to carry out the recognition process in 3D shapes, the proposed framework uses a 2D CNN, taking sections passing through the center of the shape and parallel with each of the Cartesian coordinate system planes ( x − y , y − z , x − z ) as inputs.…”

“…The results of the categorizations, can be considered as features used by different viewpoints of process planning without need to define them explicitly; in fact, categories considered in the examples were intended to be simple cases, for the illustrative purpose of these examples. 2D CNN’s are one of the most powerful tools in the categorization of images ( Tong, 2018 ) and by enriching their training data and some changes in their architecture, they are capable of categorizing a vast variety of images.…”