Skip to content

This presentation provides an overview of Machine Learning, including its history, types, algorithms, and applications. It covers topics such as supervised and unsupervised learning,linear regression, classification, clustering, model selection, evaluation, and deployment.

Notifications You must be signed in to change notification settings

SAhmadrezaAnaami/basics-of-AI-and-ML

Repository files navigation

basics-of-AI-and-ML

This presentation provides an overview of Machine Learning, including its history, types, algorithms, and applications. It covers topics such as supervised and unsupervised learning,linear regression, classification, clustering, model selection, evaluation, and deployment.

This repo is a great resource for beginners and experienced practitioners alike to learn about this important field.

this presentation created using a platform called marp check https://github.com/marp-team/marp

Created by Ahmadreza Anaami


AI: From the pages of science, to the future of our lives.

by Ahmadreza Anaami


Goals

In this lab you will:

  • have a brief introduction on AI and ML
    • basic concept of machine learning
  • main algorithm
    • supervised
      • regression
      • classification
    • unsupervised
  • Learn to implement linear regression

bg right:44%


bg 93%


machine learning

Subfield of AI in order to make intelligent machine

Field of study that gives the computers the ability to learn without explicitly programmed

  • Arthur samuel 1959

Supervised learning

  • learn from being given right answers
    • correct pairs of input (x) and output (y)

bg left:44%


input output application
email spam?(0,1)
audio text transcript
English Spanish
ad-User click?(0,1)
image-radar position of other car
image of phone defect?(0,1)

bg right:44%


Regression

predict a number from infinitely many possible outputs

bg right:55% 100%


Classification

many inputs some outputs called categories

Size diagnosis
6 1
8 1
2 0
5 0
1 0
7 1
5.6 1
12 1
3.5 0

bg right:55% 100%


bg right:100% 100%

bg left:100%


Recap

bg

In every living man a child is hidden that wants to play


bg right:100% 100%


Unsupervised learning

Find something intreating in (unlabeled) Data

  • Clustering
  • anomaly detection
  • Dimensionality reduction

bg left:60%


bg center: 100%


bg right:100% 100%

bg left:100%


bg left: 100%


linear regression

bg right:55% 100%

$$f_{w,b}(x) = wx + b$$

w , b > parameters , weight x > single feature


bg right:55% 100%

Cost function

$$J(w,b) = \frac{1}{2m} \sum\limits_{i = 0}^{m-1} (f_{w,b}(x^{(i)}) - y^{(i)})^2$$

where

$$yHat = f_{w,b}(x^{(i)}) = wx^{(i)} + b $$

m = number of example


bg center:70% 100%

Example

choose w to minimize J(w)

$$yHat = f_{w}(x^{(i)}) = wx^{(i)} $$

$$J(w) = \frac{1}{2m} \sum\limits_{i = 0}^{m-1} (f_{w}(x^{(i)}) - y^{(i)})^2$$


bg center:100% 100%

bg left:100% 100%


Gradient descent

a more systematic way to minimize j(w,b)

$$\newline\begin{align*} \text{repeat}&\text{ until convergence:} ; \lbrace \newline ; w &= w - \alpha \frac{\partial }{\partial w} J(w,b) ; \newline \newline b &= b - \alpha \frac{\partial }{\partial b} J(w,b) \newline \rbrace \end{align*}$$

bg left:50% 100%


Derivative of the Cost func

$$J(w,b) = \frac{1}{2m} \sum\limits_{i = 0}^{m-1} (f_{w,b}(x^{(i)}) - y^{(i)})^2$$

$$ \begin{align} \frac{\partial }{\partial w}J(w,b) &= \frac{1}{m} \sum\limits_{i = 0}^{m-1} (f_{w,b}(x^{(i)}) - y^{(i)})x^{(i)} \\ \frac{\partial }{\partial b}J(w,b) &= \frac{1}{m} \sum\limits_{i = 0}^{m-1} (f_{w,b}(x^{(i)}) - y^{(i)}) \\ \end{align} $$


optional

bg right:50% 100%

if alpha is too small :
if alpha is too big :

$$\alpha \frac{\partial }{\partial w} J(w,b)$$

calculate Derivative : ☻

Lets get our hands dirty

  • step 1
    • read our data
  • step 2
    • write our functions
      • calculate yHat
      • calculate cost
      • calculate gradient
      • implement gradient descent
  • step 3
    • congratulations

bg left:44%

About

This presentation provides an overview of Machine Learning, including its history, types, algorithms, and applications. It covers topics such as supervised and unsupervised learning,linear regression, classification, clustering, model selection, evaluation, and deployment.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published