0.25
0.5
0.75
1.25
1.5
1.75
2
Machine Learning
Published on Jul 27, 201736133 Views
Related categories
Chapter list
Introduction to Machine Learning00:00
Outline00:16
Types of machine learning problems00:35
Supervised learning00:46
Example: Face detection and recognition02:02
Reinforcement learning02:23
Example: TD-Gammon (Tesauro, 1990-1995)02:54
Unsupervised learning04:04
Example: Oncology (Alizadeh et al.)04:21
Example: A data set05:06
Data (continued)05:41
Terminology06:21
More formally06:41
Supervised learning problem07:01
Steps to solving a supervised learning problem07:49
Example: What hypothesis class should we pick?08:38
Linear hypothesis09:10
Error minimization!09:51
Least mean squares (LMS)10:00
Steps to solving a supervised learning problem10:37
Notation reminder10:40
A bit of algebra11:10
The solution11:38
Example: Data and best linear hypothesis12:19
Linear regression summary12:23
Linear function approximation in general12:26
Linear models in general13:39
Remarks14:23
Order-3 fit16:18
Order-4 fit16:22
Order-5 fit16:24
Order-6 fit16:26
Order-7 fit16:29
Order-9 fit16:39
Order-8 fit17:18
Order-2 fit17:42
Overfitting18:15
Overfitting and underfitting18:23
Overfitting more formally20:40
Typical overfitting plot21:59
Cross-validation23:51
The anatomy of the error of an estimator24:52
Bias-variance analysis26:10
Recall: Statistics 10127:13
The variance lemma27:33
Error decomposition30:33
Bias-variance decomposition (2)32:03
Bias-variance decomposition32:27
Bias-variance trade-off33:38
More on overfitting34:16
Coming back to mean-squared error function...35:21
A probabilistic assumption35:55
Bayes theorem in learning36:35
Choosing hypotheses37:59
Maximum likelihood estimation39:02
The log trick40:30
Maximum likelihood for regression40:48
Applying the log trick41:07
Maximum likelihood hypothesis for least-squares estimators44:23
A graphical representation for the data generation44:31
Regularization47:02
Regularization for linear models48:00
What L2 regularization does50:45
Visualizing regularization (2 parameters)51:02
Pros and cons of L2 regularization53:01
L1 Regularization for linear models53:53
Solving L1 regularization54:56
Visualizing L1 regularization55:27
Pros and cons of L1 regularization55:29
Example of L1 vs L2 effect56:07
Bayesian view of regularization - 156:50
Bayesian view of regularization - 259:05
What does the Bayesian view give us? - 101:02:36
What does the Bayesian view give us? - 201:03:46
What does the Bayesian view give us? - 301:04:44
Logistic regression01:13:26
The cross-entropy error function01:15:53
Cross-entropy error surface for logistic function01:17:22
Gradient descent01:18:03
Example gradient descent traces01:18:46
Gradient descent algorithm01:19:13
Maximization procedure: Gradient ascent01:20:18
Another algorithm for optimization01:21:15
Application to machine learning01:21:51
Second-order methods: Multivariate setting01:22:12
Which method is better?01:23:11
Newton-Raphson for logistic regression01:23:47
Regularization for logistic regression01:24:09
Probabilistic view of logistic regression01:25:10
Recap01:25:17