Machine Learning (Coursera, Andrew Ng) Flashcards
Usages for Machine Learning?
Applications cant program by hand
Examples of Database mining?
Making sense of web-click data, medical records
Making sense of web-click data is about?
Data Mining
Making sense of medical records is about?
Data Mining
Amazon/Netflix recommendations is an example of?
Self-customizing programs
Field of study that gives computers the ability to learn without being explicitly programmed?
Machine Learning
A computer program is said to learn from experience E with respect to some task T and some performance measure P, if its performance on T, measured by P, improves with experience E?
Well-posed Learning Problem
What is Machine Learning?
Field of study that gives computers the ability to learn without being explicitly programmed
What is Well-posed Learning Problem?
A computer program is said to learn from experience E with respect to some task T and some performance measure P, if its performance on T, measured by P, improves with experience E
Two broad classifications in ML?
Supervised Learning - Unsupervised Learning
What problems Supervised Learning solves?
Given the data to learn on - can we predict the future results?
What problems Unsupervised Learning solves?
Given a dataset - can we find some structure?
The ML Problems?
Regression, Classification, Clustering
Regression?
Predict continuous value output
Classification?
Predict discrete value output
Clustering?
Grouping the set of inputs together
To what ML Problems corresponds - Price Prediction?
Regression
In Regression problem what does m stand for?
Number of training examples
In Regression problem what does x stand for?
Input feature
In Regression problem what does y stand for?
Output variable
In Regression problem what does h stand for?
Hypothesis function
Examples for Regression Application?
Price prediction
Formula for linear regression?
h(x) = theta0 + theta1 * x
Purpose of cost function?
Estimate correctness of our hypothesis function
Squared error cost function?
https://www.codecogs.com/eqnedit.php?latex=J(%5Ctheta_0%2C%20%5Ctheta_1)%20%3D%20%5Cfrac%7B1%7D%7B2m%7D%20%5Csum_%7Bi%3D1%7D%5Em(%20%5Chat%7By%7D_i%20-%20y_i%20)%5E2%20%3D%20%5Cfrac%7B1%7D%7B2m%7D%20%5Csum_%7Bi%3D1%7D%5Em(%20h_%5Ctheta%20(x_i)%20-%20y_i)%5E2
How we choose parameters for linear regression?
In order to minimize the difference between hypothesis and real value
What is Gradient Descent?
Minimization Algorithm
What does “Batch Gradient Descent” mean?
All training examples are analyzed on each step
How does Gradient Descent Algorithm look like?
repeat until convergence {
} where j=0,1 represents the feature index number. (Simultaneous assignment)
What happens if learning rate of Gradient Descent is too small?
It might take a lot of steps before convergence
What happens if learning rate of Gradient Descent is too big?
It may fail to converge, or even diverge
Can Gradient Descent converge with fixed learning rate?
Yes
Why Gradient Descent converges with fixed learning rate?
Because derivative is changing as we approach the minimum