![信息论、推理与学习算法 - [英] David J.C. MacKay](https://static.book345.com/covers/s/9787040196412.jpg)
书籍介绍
这不是一本教你速成机器学习的技术手册，而是一部把信息论从通信工程搬进概率推理世界的思想著作。作者以
AI导读
核心看点
- 统一框架融合信息论、贝叶斯推理与机器学习
- 涵盖数据压缩、信道编码、神经网络等核心主题
- 提供丰富例题与近400道习题，含详细解答
读者共识
- 逻辑自洽编排流畅，第四部分尤为适合单独研读
- 公式推导严谨，对理解变分贝叶斯极具参考价值
- 中文译本口语感强，部分读者认为翻译质量一般
精彩摘录
- "Probabilities can be used in two ways. 1. Probabilities can describe frequencies of outcomes in random experiments 2. Probabilities can also be used, more generally, to describe degrees of belief in propositions that do not involve random variables This more general use of probability to quantify be"
- "The details of the other possible outcomes and their probabilities are irrelevant. All that matters is the probability of the outcome that actually happened (here, that the ball drawn was black) given the dierent hypotheses. We need only to know the likelihood, i.e., how the probability of the data "
- "Entropy is maximized if p is uniform: [;H(X) \leq log(|A_X|);] with equality iff [;p_i = 1/|A_X|;] for all i."
- "a disk drive is an example of communication, which doesn't have to involve information going from one place to another. When we write a file on a disk drive, we'll read it off in the same location - but at a later time."
目录
用户评论
【翻过】适合做参考书。国内不少学生聪明勤奋有热情，可惜所在学校没有好导师指导，很容易误入歧途。建议：1）一定要抱大腿。找一下全国各地有哪些大牛，挤破头也要挤进他们的实验室。我当年本科就休学一年去了中科院。2）读读下面这本书。别的书教你知识，这本书教你思考。学习一流学者的思维过程，比具体知识更重要。——路遥_机器学习@新浪微博
概率与推理， 神经网络， 稀疏图码
后面比较精彩 尤其是基因
有提到模型和数据信息量等问题
对于信息传递的讲的非常清楚。
老吐槽翻译有什么意思，原文的口语感本来就很强。概括精炼，第四部分适合单独成书，逻辑自洽、内容编排流畅、小而精太适合自学了。但跳过习题估计是很难读下去的，不适合做工具书。
好的教材真是比浪费时间的科普强多了
有习题还有解答。涉及知识面很广，例子很犀利。好书推荐。
下载
收藏