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。

To Build Truly Intelligent Machines, Teach Them Cause and Effect

33 点作者 DmenshunlAnlsis大约 7 年前

2 条评论

YeGoblynQueenne大约 7 年前
&gt;&gt; Mathematics has not developed the asymmetric language required to capture our understanding that if X causes Y that does not mean that Y causes X.<p>There we go again. These assertions by Pearl discredit him. Yes, mathematics can capture this perfectly well, it&#x27;s the implication relation in a nutshell: X → Y means that X causes Y but we don&#x27;t know whether Y causes X.<p>I guess, strictly speaking, implication doesn&#x27;t say anything specific about causality, so you could conceivably claim that A → B (&quot;If A, then B&quot;) is not a causal relation, because even when B always follows A we can&#x27;t be assured that A is the cause of B (e.g. two alarms might go off one after the other always), but that is really splitting hairs. The semantics of implication are broad enough that they can cover both strict causality relations _and many more besides_. The intended meaning can always be made explicit in language where it&#x27;s not clear from the context ...and that&#x27;s what Pearl is most likely complaining about. He basically wants a stricter interpretation that can only cover causal relations so that the meaning doesn&#x27;t depend on the context. Because!<p>The whole point here is that Pearl wants his causal reasoning framework to be accepted by everyone, even if it&#x27;s not really offereing anything radically new.
评论 #17081889 未加载
评论 #17081554 未加载
drallison大约 7 年前
Correlation does not imply Causality. Pearl&#x27;s work shows that it is sometimes possible to reason and make inferences about causal relationships in data even when a controlled experiment is not possible and sometimes not. Machine learning algorithms that differentiate between known causal relations from correlative relations should perform better. The amazing thing is how good an approximation to causality correlation may be.<p>Interested in persuing this further? See <a href="http:&#x2F;&#x2F;www.michaelnielsen.org&#x2F;ddi&#x2F;if-correlation-doesnt-imply-causation-then-what-does&#x2F;" rel="nofollow">http:&#x2F;&#x2F;www.michaelnielsen.org&#x2F;ddi&#x2F;if-correlation-doesnt-impl...</a> and, I suppose, read Pearl&#x27;s latest book.