This post was published in 2021-11-20. Obviously, expired content is less useful to users if it has already pasted its expiration date.
Table of Contents
从Chrome换到了Edge一段时间,使用感受
从Chrome换到了Edge一段时间,使用感受:
1,Edge的“自动冻结标签页”确实非常省电。
2,Edge的下载机制非常不适应,简直就是弱智。
3,Edge 历史记录 的UI设计真的非常难看。为什么就不能像Chrome那样搞个全屏呢?
学习了一波pycURL
学习了一波pycURL,试图用来代替requests,看起来性能会好一些:
🔗 [python - pycurl and SSL cert - Stack Overflow] https://stackoverflow.com/questions/8332643/pycurl-and-ssl-cert
🔗 [Pycurl 的属性和方法 – 均益个人博客] https://junyiseo.com/python/607.html
🔗 [python - Get header values of reply using pycurl - Stack Overflow] https://stackoverflow.com/questions/15641080/get-header-values-of-reply-using-pycurl,配合🔗 [python - StringIO in Python3 - Stack Overflow] https://stackoverflow.com/questions/11914472/stringio-in-python3
成果:
成功用pycURL替换了preload.py的requests,可以非常明显的看到:preload.py运行的cpu消耗从原来的~15%降低到~5%,运行整个preload.py的时间降低了3~4s,但页面平均生成时间并没有明显进步。
(当然, 页面平均生成时间 本身就是个玩具指标,看看就好)
猜测页面平均生成时间没有明显进步的原因:当前preload.py只开了2个线程,大多数时候都是“php-fpm生产html(php-fpm高负载)-> python3接受并解析很长的String(python3高负载)->php-fpm生产html(php-fpm高负载)..... ”这样的流程,总的来说就是python3的高cpu使用时机和php-fpm的高cpu使用时机大部分都错开了。
贝叶斯网络/BN/Bayesian Network
2022-08更新:下面的这些笔记仅仅粘贴了一些链接,更详细的笔记和推导见:🔗 [贝叶斯网络 - Truxton's blog] https://truxton2blog.com/bayesian-network-introduce/
preview preview2 preview3 preview4置顶一个bayesian net的实际应用,快速上手了解:https://www.zhihu.com/question/28006799/answer/38996563
一个中文slide:preview 🔗 [bnet_slides.pdf] http://staff.ustc.edu.cn/~jianmin/lecture/AI2014/bnet_slides.pdf
🔗 [Bayesian network - Wikipedia] https://en.wikipedia.org/wiki/Bayesian_network
最简单的一个贝叶斯网络例子和例题:
🔗 [*PowerPoint Presentation] https://web.engr.oregonstate.edu/~tgd/talks/dietterich-chile-jcc2012-v1.pdf
🔗 [How important are probabilistic graphical models for machine learning? - Quora] https://www.quora.com/How-important-are-probabilistic-graphical-models-for-machine-learning
🔗 [贝叶斯网络(Bayesian network))简介(PRML第8.1节总结)概率图模型(Graphical models) - joey周琦 - 博客园] https://www.cnblogs.com/Dzhouqi/p/3204353.html
🔗 [条件独立(conditional independence) 结合贝叶斯网络(Bayesian network) 概率有向图 (PRML8.2总结) - joey周琦 - 博客园] https://www.cnblogs.com/Dzhouqi/p/3204481.html
🔗 [[概率图模型]1_贝叶斯网络的模型表示_哔哩哔哩_bilibili] https://www.bilibili.com/video/BV1AQ4y1P7Da
🔗 [Reasoning Under Uncertainty: Variable Elimination Example] https://www.cs.ubc.ca/~kevinlb/teaching/cs322%20-%202006-7/Lectures/lect29.pdf
这个链接里面有一道经常拿来复习的例题(* 这道例题在笔记「贝叶斯网络」中也提到过):🔗 [Bayesian Networks - Exact Inference by Variable Elimination] https://www.cs.upc.edu/~larrosa/MEI-CSI-files/BN/2-BN-VE.pdf