This shows you the differences between two versions of the page.

Link to this comparison view

feedback_network [2017/08/20 10:52] (current)
Line 1: Line 1:
 +http://​feedbacknet.stanford.edu/ ​
 +An alternative that can achieve
 +the same goal is a feedback based approach in which the
 +representation is formed in an iterative manner based on a
 +feedback received from previous iteration’s output.
 +We establish that a feedback based approach has several
 +fundamental advantages over feedforward:​ it enables
 +making early predictions at the query time, its output naturally
 +conforms to a hierarchical structure in the label space
 +(e.g. a taxonomy), and it provides a new basis for Curriculum
 +Learning.  ​
 + ​http://​feedbacknet.stanford.edu/​
 +http://​jmlr.org/​proceedings/​papers/​v38/​lee15a.pdf ​ Deeply Supervised Networks
 +https://​arxiv.org/​pdf/​1708.04483v1.pdf Learning with Rethinking: Recurrently Improving
 +Convolutional Neural Networks through Feedback
 +https://​arxiv.org/​abs/​1607.06854 Unsupervised Learning from Continuous Video in a Scalable Predictive Recurrent Network