We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

stat.ML

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo

Statistics > Machine Learning

Title: Deep Layer-wise Networks Have Closed-Form Weights

Abstract: There is currently a debate within the neuroscience community over the likelihood of the brain performing backpropagation (BP). To better mimic the brain, training a network $\textit{one layer at a time}$ with only a "single forward pass" has been proposed as an alternative to bypass BP; we refer to these networks as "layer-wise" networks. We continue the work on layer-wise networks by answering two outstanding questions. First, $\textit{do they have a closed-form solution?}$ Second, $\textit{how do we know when to stop adding more layers?}$ This work proves that the kernel Mean Embedding is the closed-form weight that achieves the network global optimum while driving these networks to converge towards a highly desirable kernel for classification; we call it the $\textit{Neural Indicator Kernel}$.
Comments: This version will be published in AIStats 2022
Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG)
Cite as: arXiv:2006.08539 [stat.ML]
  (or arXiv:2006.08539v6 [stat.ML] for this version)

Submission history

From: Chieh T Wu [view email]
[v1] Mon, 15 Jun 2020 16:48:03 GMT (5729kb,D)
[v2] Wed, 4 Nov 2020 18:40:56 GMT (5723kb,D)
[v3] Wed, 2 Dec 2020 16:38:13 GMT (5725kb,D)
[v4] Thu, 3 Dec 2020 14:57:29 GMT (5725kb,D)
[v5] Mon, 7 Feb 2022 17:13:16 GMT (6669kb,D)
[v6] Wed, 9 Feb 2022 17:42:52 GMT (6595kb,D)

Link back to: arXiv, form interface, contact.