<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Avoiding Implementation Pitfalls of "Matrix Capsules with EM Routing" by Hinton et al · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/1907.00652/</link><description>Collected HN comments and submissions linking this paper.</description><language>en</language><lastBuildDate>Mon, 02 Nov 2026 11:16:22 GMT</lastBuildDate><atom:link href="https://hackerpaper.kaipfister.de/p/1907.00652/feed.xml" rel="self" type="application/rss+xml" /><item><title>Avoiding Implementation Pitfalls of "Matrix Capsules with EM Routing" by Hinton et al — mentioned by fheinsen</title><link>https://news.ycombinator.com/item?id=21412532</link><guid isPermaLink="false">urn:hackerpapers:mention:1907.00652:21412532</guid><pubDate>Thu, 31 Oct 2019 19:16:12 GMT</pubDate><description>&lt;p&gt;Great questions. Happy to answer them here. First of all, this work builds on Hinton et al.’s second paper, the one about EM routing of matrix capsules, from last year: https://ai.google/research/pubs/pub46653 This work is only minimally related to his previous paper (Sabour et al.&amp;#x27;s paper) from two years ago! RESPONSES TO #1: * The same algorithm also achieves SOTA in another domain, natural language. Same code. I think it’s significant that the same code, without change, produces SOTA in two domains. See the README and tables 3 and 4 in the draft paper. Don&amp;#x27;t you think this is significant?…&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/1907.00652/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.LG</category></item></channel></rss>
