TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Geometric Understanding of Deep Learning (2018)

164 pointsby yoquanover 6 years ago

9 comments

cs702over 6 years ago
Wow. As far as I know, this is the first time anyone reputable[a] has claimed to <i>show</i> (!) that the &quot;manifold hypothesis&quot; is the fundamental principle that makes deep learning work, as has long been believed:<p><pre><code> &quot;In this work, we give a geometric view to understand deep learning: we show that the fundamental principle attributing to the success is the manifold structure in data, namely natural high dimensional data concentrates close to a low-dimensional manifold, deep learning learns the manifold and the probability distribution on it.&quot; </code></pre> Moreover, the authors also claim to have come up with a way of measuring how hard it is for any deep neural net (of fixed size) to learn a parametric representation of a particular lower-dimensional manifold embedded in some higher-dimensional space:<p><pre><code> &quot;We further introduce the concepts of rectified linear complexity for deep neural network measuring its learning capability, rectified linear complexity of an embedding manifold describing the difficulty to be learned. Then we show for any deep neural network with fixed architecture, there exists a manifold that cannot be learned by the network.&quot; </code></pre> Finally, the authors also propose a novel way to control the probability distribution in the latent space. I&#x27;m curious to see how their method compares and relates to recent work, e.g., with discrete and continuous normalizing flows:<p><pre><code> &quot;...we propose to apply optimal mass transportation theory to control the probability distribution in the latent space.&quot; </code></pre> This is <i>not</i> going to be a light read...<p>--<p>[a] One of the authors, Shing-Tung Yau, is a Fields medalist: <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=18987219" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=18987219</a>
评论 #18990691 未加载
评论 #18991122 未加载
lovelearningover 6 years ago
[An OT question as somebody not familiar with the academic world] Two of the authors are in a Chinese university. Two of them in different departments in a US university. In general, how does this kind of intercontinental collaboration start, and how do they progress? How are roles defined when multiple people are involved in a theoretical paper like this? Are there some tools that help with collaborative paper writing?
评论 #18989068 未加载
评论 #18989687 未加载
jesuslopover 6 years ago
Possibly less sophisticatedly I think of them as a sandwich of affine maps and nonlinear isotropies (as those giving irregular rings in tree trunks). The affinities are represented nicely in GL(n+1) with a homogenous coordinates trick related to neuron biases. A question would be if there&#x27;s something interesting to say about the interactions of the affinities and isotropies in group theoretic terms (which I dunno).
评论 #18989539 未加载
yoquanover 6 years ago
I&#x27;m reading this and just realized one author is the famous Field medalist, Shing-Tung Yau :-)
评论 #18988834 未加载
throwawaymathover 6 years ago
<i>&gt; ...we show that the fundamental principle attributing to the success is the manifold structure in data...<p>&gt; Then we show for any deep neural network with fixed architecture, there exists a manifold that cannot be learned by the network.</i><p>I&#x27;d venture a guess that you can extend this result to show that, for any deep neural network with fixed architecture, there exists an adversarial manifold it must be vulnerable to.<p>In other words not only is there a manifold the neural network <i>cannot</i> learn, but there is also a manifold it <i>will</i> learn, but incorrectly.
crimsonalucardover 6 years ago
Anybody know what I should study in order to understand this research paper?
评论 #18991238 未加载
ttfleeover 6 years ago
Another interesting paper on optimal transportation and GAN: <a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1710.05488" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1710.05488</a>
twicover 6 years ago
I know next to nothing about deep learning. But this geometric interpretation really reminds of of the way self-organising maps work. Is there a real connection there, or is that superficial?
评论 #18993108 未加载
heinrichfover 6 years ago
(uploaded to the arXiv in May 2018)
评论 #18988852 未加载