<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Neural Arithmetic Expression Calculator · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/1809.08590/</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/1809.08590/feed.xml" rel="self" type="application/rss+xml" /><item><title>Neural Arithmetic Expression Calculator — mentioned by olooney</title><link>https://news.ycombinator.com/item?id=20669986</link><guid isPermaLink="false">urn:hackerpapers:mention:1809.08590:20669986</guid><pubDate>Sun, 11 Aug 2019 19:39:22 GMT</pubDate><description>&lt;p&gt;&amp;gt; You’re not going to teach your dog relativity with back propagation What do you mean? RNN&amp;#x27;s can be taught to do math using BP[1]. Heck, you can do a simple version yourself in less than an hour[2]. They can&amp;#x27;t do tensor calculus yet (AFAIK) but there doesn&amp;#x27;t seem be any reason why BP in particular would be the stumbling block; the difficulty is finding the right representation. &amp;gt; yet we treat BP like it was going to solve all the problems in AI Except for reinforcement learning which does not use BP to solve the Bellma equations[3], AlphaZero which uses minimax[4], AutoML based on…&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/1809.08590/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.AI</category></item></channel></rss>
