Scribd
Upload
3 views26 pages

AAI Lecture 9 SP 25

The document provides an introduction to machine learning, distinguishing it from artificial intelligence (AI) and deep learning (DL) based on training set requirements and feature determination. It outlines various types of learning, including supervised, unsupervised, and reinforcement learning, along with their applications and challenges such as overfitting and dimensionality. The document emphasizes the importance of data, features, and evaluation metrics in the machine learning process.

Uploaded by

i228791
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
3 views26 pages

AAI Lecture 9 SP 25

The document provides an introduction to machine learning, distinguishing it from artificial intelligence (AI) and deep learning (DL) based on training set requirements and feature determination. It outlines various types of learning, including supervised, unsupervised, and reinforcement learning, along with their applications and challenges such as overfitting and dimensionality. The document emphasizes the importance of data, features, and evaluation metrics in the machine learning process.

Uploaded by

i228791
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 26

Introduction to Machine

Learning

1
AI vs Machine Learning vs Deep Learning
Summary

Scheme Training Set required Pre-Determined Features


required

AI No Yes

ML Yes Yes

DL Yes No
Timeline
What is learning?

- An agent is learning if it improves performance after making


observations about the world

- Machine learning: When a computer learns from data

5
Why machine learning?
• The task: write a program that, given a 28x28 grayscale image of a digit,
outputs the number in the image

• Image: digits from the MNIST data set


(http://yann.lecun.com/exdb/mnist/)
6
Approaches
• Approach 1: try to write a program by
hand that uses your a priori knowledge
about what images look like to determine
what number they are 8

• Approach 2: (the machine learning


approach) collect a large volume of
images and their corresponding numbers, 5
let the “write its own program” to map
from these images to their corresponding
number

7
Learning Problem
• Learning = Improving with experience over some task
• Improve over task T,
• With respect to performance measure P,
• Based on experience E.
• Example
• T = Play checkers
• P = % of games won in a tournament
• E = opportunity to play against itself

8
Machine Learning
Learning = Improving with experience over some task

A computer program is said to learn from experience


E with respect to some task T and performance
measure P, if its performance at task T, as measured
by P, improves with experience E

9
Types of learning
• Supervised learning: learn from (input, output) pairs

• Unsupervised learning: Detecting patterns from inputs alone (for


e.g. clustering)

• Reinforcement learning: Learn from a series of rewards and


punishments

10
Supervised Learning
• Supervised Learning
• Classification
• Regression
ML (Reflex) Models
Classification
Features
• Use fewer features if possible
• Use features that differentiate classes well
• Character recognition example
• Good features: aspect ratio, presence of loops
• Bad features: number of black pixels, number of connected components

21
Less is More
The Curse of Dimensionality
• Learning from a high-dimensional feature space requires an enormous amount
of training to ensure that there are several samples with each combination of
values.
• With a fixed number of training instances, the predictive power reduces as
the dimensionality increases.

• As a counter-measure, many dimensionality reduction techniques have been


proposed,
• when done properly, the properties or structures of the objects can be well preserved
even in the lower dimensions.
• Naively applying dimensionality reduction can lead to incorrect results.
Training, Validation and Testing
Important Concepts
• Data: labeled instances (e.g. emails marked spam/ham)
• Training set
• Held out set
• Test set

• Features: attribute-value pairs which characterize each x Training


Data
• Experimentation cycle
• Learn parameters (e.g. model probabilities) on training set
• (Tune hyperparameters on held-out set)
• Compute accuracy of test set
• Very important: never “peek” at the test set!

• Evaluation (many metrics possible, e.g. accuracy) Held-Out


• Accuracy: fraction of instances predicted correctly Data

• Overfitting and generalization


• Want a classifier which does well on test data Test
• Overfitting: fitting the training data very closely, but not Data
generalizing well
Unsupervised Learning
• No labelled training set required.

• Example (Clustering):
• Imagine you have 50 students and for each student you know his/her
marks out of 100. Divide these 50 students in good, average and worst
categories based upon their marks.
• Label in this case is one of the three strings (“good”, “average” or “bad”).
Here you have a training set (a list of marks of those 50 students) but you
don’t know the label of each of that student (whether that student
belongs to a “good” category or one of the other two. That is what you
have to find out!
Reinforcement Learning
• In reinforcement learning the agent learns from a series of
reinforcements—rewards or punishments.
• For example, the lack of a tip at the end of the journey gives the
taxi agent an indication that it did something wrong. The two
points for a win at the end of a chess game tells the agent it did
something right.

You might also like