🪴 Freeman's Second Brain

Search

SearchSearch

Supervise Learning Guide

Aug 27, 2024, 1 min read

Supervise Learning Guide

Definition:
Supervised learning models are a type of machine learning algorithm that involves training a model to predict an output based on input data, where the output is known for a set of labeled examples. Supervised learning is used in a variety of applications, such as image and speech recognition, natural language processing, and recommendation systems.
The choice of model depends on the problem at hand and the type of data available. It is important to evaluate the performance of the model on a separate test set to ensure that it generalizes well to new data. **Regularization techniques** can also be used to prevent overfitting and improve the performance of the model.

Train-Validation-Test sets Scaling

Models:

Linear Regression Guide

Linear Regression in ML

Linear Regression in ML

Review of Linear Model:

  1. are matrices or vectors.
  2. A row or column of a matrix will use a subscript:
    • is a scalar, the ith row and jth column of
    • . is a vector, the ith row of
    • is a vector, the jth column of
  3. Parameters will typically be Greek symbols:
  4. Estimates will be Greek symbols wearing a hat:
  5. is the probability that given , where:
    • X, Y are random variables
  6. represents either a loss function, or the likelihood function
    • represent the log-likelihood function

X(n by p+1), features with columns of list Y(n by 1), target variable B(p+1 by 1) coefficient, parameters Y = XB + E

Training linear model means finding optimal coefficients :

Finding optimal amounts to find vector that minimizes the loss function , which is the sum of squared error:

which is equivalent to:

Finding

by taking the derivative of our loss function and setting it equal to zero

TAGS

Link to original

Regluarization

Regluarization

Notes and Ideas:

Addjusted cost function

M(w): Model Error

R(w): Function of Estimated Parameter(s)

: regularization strength parameter, adds a penalty proportional to the size of the estimated model parameters, or a function of the parameter

Link to original
Regluarization performs Feature Selection by shrinking the contribution of features.

For L1-regularization (Lasso Regression), this is accomplished by driving some coefficients to zero.

Reducing the number of features can prevent overfitting.

Key words:

TAGS

Link to original

Key words:

TAGS

Graph View

Backlinks