TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

What's wrong with convolutional nets? [video]

113 点作者 drewvolpe超过 10 年前

6 条评论

jfoutz超过 10 年前
Short, massively simplified, summary,<p>Vast improvement in unsupervised learning.<p>Walks through biases humans have in object recognition. Explains new structures to add on top of neural networks.<p>Basic backprop on MNIST with 30,000 labeled examples gives 1.7% error rate. Hinton&#x27;s approach gets the same results with 25 examples.
评论 #8784483 未加载
robg超过 10 年前
One of the few speakers I saw as an academic and said to myself &quot;This guy is a time traveler. I need to really pay attention to what he&#x27;s saying.&quot;<p>That was in 2001. A decade+ later and we&#x27;re still trying to catch up.
评论 #8786864 未加载
评论 #8784262 未加载
评论 #8787452 未加载
TrainedMonkey超过 10 年前
Video buffers every few seconds for me, are there any mirrors?<p>Edit: found download link after getting some caffeine <a href="http://techtv.mit.edu/videos/30698-what-s-wrong-with-convolutional-nets/download.source" rel="nofollow">http:&#x2F;&#x2F;techtv.mit.edu&#x2F;videos&#x2F;30698-what-s-wrong-with-convolu...</a><p>Edit 2: Also, there is a class on Coursera where he shares much knowledge and intuition about neural networks. I personally took it and it was excellent. Unfortunately there are no repeating sessions, but even archived material is a goldmine. <a href="https://class.coursera.org/neuralnets-2012-001" rel="nofollow">https:&#x2F;&#x2F;class.coursera.org&#x2F;neuralnets-2012-001</a>
评论 #8798666 未加载
discardorama超过 10 年前
If I&#x27;m understanding some of the criticism, it is basically that we don&#x27;t think &quot;that&#x27;s how our brain does it&quot; (when it comes to convnets, pooling, etc.). But this begs the question: do we _have_ to duplicate what goes on in the brain? For example: we really don&#x27;t know how the brain does multiplication; and if I were to guess, I&#x27;d say it has nothing to do with binary arithmetic and shift registers. But yet we&#x27;ve been able to teach a computer how to do it, using an algorithm that has (in all probability) no relation to how our brains work. So maybe convnets, pooling and other hacks work, just not how the brain works?
akyu超过 10 年前
A lot of these sound like ideas that Jeff Hawkins has been pushing with his Cortical Learning Algorithm
评论 #8786058 未加载
评论 #8785802 未加载
jacobn超过 10 年前
Does anyone have a link to the presentation itself (e.g. ppt&#x2F;pdf)?