Lecture 11 | Machine Learning (Stanford)
Why Cross Validated Stack Exchange recommends it
Cross Validated users link it when they need Andrew Ng's practical guidance on getting ML to work, such as learning-curve diagnostics and comparing SVM against logistic regression.
Recommended by 1 people across 1 different months, 2015 to 2017.
What it is
Lecture 11 of Stanford's CS 229 Machine Learning course, taught by Andrew Ng in 2008. It covers Bayesian statistics, regularization, online learning, and practical advice on applying machine learning algorithms, including diagnosing classifiers with learning curves.
What people said
Recommends d0rmLife · 2015-10 · Cross Validated Stack Exchange · on “Lecture video of how to get machine learning to work in practice”
Andrew Ng has lectures on ML on Coursera [LINK] , which includes a section of suggestions for applications. There are also Stanford lecture videos on [this video] . I have linked the video you seek.
Cites Anton · 2016-07 · Cross Validated Stack Exchange · on “Doing cross-validation when diagnosing a classifier through learning curves”
I have a theoretical question on the correct way to make learning curves to diagnose a classifier. To see a generic example of these curves one can refer to this [this video] (min 34 onward) lecture by Andrew Ng or this [LINK] example from scikit. There is sparse information on the topic elsewhere -- that is my reason to ask experts here. My understanding on the matter can be briefly summarized in the procedure below. Make a train/test split of the data set. The test set is held constant over the next steps. Derive a training set subsample of size N and train your classifier on it. Evaluate your classifier performance on the test set you derived at step #1 to get a point for the test error curve.
Cites The Wanderer · 2017-09 · Cross Validated Stack Exchange · on “Transferring parameters learned using one model to another”
I am watching this video lecture by Andrew NG on "Getting Machine Learning Algorithms to work in Practice" and I am stuck at the following two slides: The problem discussed is as follows: Let's say an SVM outperforms Logistic Regression. How do I know if there is a problem with my 'optimization' or 'optimization function'? [LINK] [LINK] He suggests to use $\theta_{svm}$ that were learned using SVM in the optimization function of logistic regression. Coincidentally, for both SVM and Logistic Regression, $\theta$s are the weights of linear features. And therefore, I can use $\theta$s learned using SVM into the optimization function of linear regression. Question : What if $\theta$s are not tha