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。

Deep Learning is optimization, not learning, says Intel “neuro guru”

3 点作者 sandwall超过 6 年前

2 条评论

itg超过 6 年前
Response from Yann LeCunn: <a href="https:&#x2F;&#x2F;www.facebook.com&#x2F;yann.lecun&#x2F;posts&#x2F;10155808247852143&amp;width=500%22" rel="nofollow">https:&#x2F;&#x2F;www.facebook.com&#x2F;yann.lecun&#x2F;posts&#x2F;10155808247852143&amp;...</a>
utdiscant超过 6 年前
I am by no means an expert on AI, but I find it really interesting to think about the possibility to actually getting to strong&#x2F;general AI using existing techniques (some form of deep learning &#x2F; neural network).<p>What is the really strong argument for why the current techniques with added computational power and continuous improvements will not lead to strong AI?<p>People often claim things like &quot;that is just optimization, not really intelligence&quot; or something similar - but it seems very hard to prove that what we perceive as human intelligence is more than just similar calculations.
评论 #19240510 未加载