Cross Entropy vs Cross Entropy Loss
这俩是有区别的，我一直没有注意直到我发现公式上的区别。我没有注意的原因应该是 regression 没有显式使用 cross entropy loss，而 classification 的 cross entropy loss 是对原生 cross entropy 的扩展。
最新文章
Tensorflow 2.0 Autograph and Tracing: an example
首先说下为啥需要把代码转成 graph。我觉得可能主要有两点考虑：
Huber loss / np.where
Huber loss:
Convolution is convolution; it’s NOT dot product
我真的是出离愤怒。我不知道最开始把 convolution 看做 dot product 的人是怎么想的！有 convolution 的公式不用，非要用这么蹩脚的 intuition？而且明显 convolution 和 Hadamard product 的关系更大一点呢，咋没见人提？
Terminology Recap: Method of Least Squares (and its relationship with MLE)
$$ \DeclareMathOperator*{\argmin}{argmin}
Terminology Recap: Generative Models / Discriminative Models / Frequentist Machine Learning / Bayesian Machine Learning / Supervised Learning / Unsupervised Learning / Linear Regression / Naive Bayes Classifier
1. Generative vs Discriminative
Install Oracle JDK 8 and switch between JDKs in Ubuntu 18.04
1. Check installed Open JDK
Terminology Recap: Sampling / Sample / Sample Space / Experiment / Statistical Model / Statistic / Estimator / Empirical Distribution / Resampling / CV / Jackknife / Bootstrap / Bagging / Likelihood / Estimation and Machine Learning
1. Sampling from a Probability Distribution
Ring of Sets
写这篇主要是因为我看到有些 measure 的定义用到了 $\delta$-ring.