Intro to ML
Definition
"The field of study that gives computers the ability to learn without being explicitly programmed." —Arthur Samuel
"A computer program is said to learn from experience E concerning some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E." — Tom Mitchell
Category
Supervised Learning
We have a dataset and already know the correct outputs, having the idea that there is a relationship between the input and the output. Supervised Learning is categorized into regression and classification problems.
-
Regression Problem
Given the historical data, fit a regression model (which is a trend) to predict the continuous output according to its inputs.
-
Classification Problem
Given the historical data, draw a line (or face) that separate the outputs into binary results.
Unsupervised Learning
Having data and having no idea what the result should look like, we can derive structure by clustering the data based on the relationship among input variables.
Model Representation
Elements
- Input
- Hypothesis Function: a function that maps inputs to outputs.
- Cost Function: a function used to measure the accuracy of hypothesis function.
- Optimization Method: a method that used to minimize the sum cost of the feeding dataset (e.g. Gradient Descent)
- Output
5 Elements of ML
Linear Algebra
Multiplication
Matrix vector
matrix-vector.png
Matrix matrix
matrix-matrix.png
Transpose
transpose.png
Inverse
inverse.png
Properties
Not commutative
A∗B ≠ B∗A
Associative
(A∗B)∗C = A∗(B∗C)
Reference
https://www.mathsisfun.com/algebra/matrix-multiplying.html
https://www.coursera.org/learn/machine-learning/resources/JXWWS
https://medium.com/@mahesh.chavan1997/model-representation-in-machine-learning-2caf22e1edf0
