You are on page 1of 15

ML INTRODUCTIN

ESLAM OSAMA
AMIT LEARNING
Agenda

Introduction Introduction

Where are we in CS? Where are we inCS?

Pattern recognition
Definition

Pattern recognition Computer Vision

Machine Learning
Definition What is Machine Learning
ML Workflow
Computer Vision ML Types

References
Machine Learning
What is Machine Learning
ML Workflow
ML Types

References

Contact Information

15
Introduction

Introduction

Where are we in CS? 2 Where are we inCS?

Pattern recognition
Definition
) Artificial intelligence, machine learning, and deep learning Computer Vision

Machine Learning
What is Machine Learning
ML Workflow
ML Types

References

15
Introduction

Introduction
3 Where are we inCS?

Pattern recognition
Definition
Computer Vision

What about Pattern Recognition? Machine Learning


What is Machine Learning
ML Workflow
) Deep learning for natural-language processing is pattern ML Types

recognition applied to words, sentences, and paragraphs, References

in much the same way that computer vision is pattern


recognition applied to pixels.

15
Pattern recognition

Introduction
Where are we in CS?

The assignment of a physical object or event to one of several Pattern recognition


4 Definition Computer
pre-specified categories” – Duda and Hart Vision

Machine Learning
1. A pattern is an object, process or event What is Machine Learning
ML Workflow
2. A class (or category) is a set of patterns that share ML Types

common attribute (features) usually from the same References

information source
3. During recognition (or classification) classes are
assigned to the objects.
4. A classifier is a machine that performs such task

15
Pattern recognition and our theme

Introduction
Where are we in CS?

Pattern recognition
Definition
5 Computer Vision

Machine Learning
What is Machine Learning
ML Workflow
ML Types

References

15
Machine Learning: Smart programs can
learn from examples

) In the early 90s people started realizing that a more Introduction


Where are we inCS?
powerful way to build pattern recognition algorithms is to Pattern recognition
replace an expert (knows way too much about pixels) with Definition
Computer Vision
data (which can be mined from cheap laborers). Machine Learning
6 What is Machine Learning
) So collect a bunch of face images and non-face images, ML Workflow

choose an algorithm, and wait for the computations to ML Types

References
finish.
) "Machine Learning" emphasizes that the computer
program (or machine) must do some work after it is given
data.

15
Machine Learning

to do machine learning, we need three things:


Introduction
Where are we inCS?
) Input data points—For instance, if the task is speech
Pattern recognition
recognition, these data points could be sound files of Definition
Computer Vision
people speaking. If the task is image tagging, they could Machine Learning
be pictures. 7 What is Machine Learning
ML Workflow
) Examples of the expected output —In a ML Types

speech-recognition task, these could be human-generated References

transcripts of sound files. In an image task, expected


outputs could be tags such as “dog,” “cat,” and so on.
) A way to measure whether the algorithm is doing a
good job—This is necessary in order to determine the
distance between the algorithm’s current output and its
expected output. The measurement is used as a feedback
signal to adjust the way the algorithm works. This
adjustment step is what we call learning.
15
Machine learning

Introduction
Where are we in CS?

Pattern recognition
Definition
Computer Vision

Machine Learning
What is Machine Learning
8 ML Workflow
ML Types

References

15
Machine learning

Introduction
Where are we in CS?

Pattern recognition
Definition
Computer Vision

All ML tasks can be classified in several categories, Machine Learning

the main ones are: What is Machine Learning


ML Workflow
9 ML Types
) supervised ML References

) unsupervised ML;
) reinforcementlearning.

15
Machine learning

Supervised ML Introduction
Where are we inCS?
) Supervised ML relies on data where the true label/class Pattern recognition
was indicated. This is easier to explain using an example. Definition
Computer Vision

) Example: Let us imagine that we want to teach a Machine Learning


What is Machine Learning
computer to distinguish pictures of cats and dogs. ML Workflow
10 ML Types
) Collect pictures of cats and dogs adding a tag ‘cat’ or References
‘dog’. Labelling is usually done by human annotators to
ensure a high quality of data.
) So now we know the true labels of the pictures and can
use this data to “supervise” our algorithm in learning the
right way to classify images. Once our algorithm learns
how to classify images we can use it on new data and
predict labels (‘cat’ or ‘dog’ in our case) on previously
unseen images.

15
Supervised ML : instances of supervised
learning

Introduction
Where are we inCS?

Pattern recognition
Definition
Computer Vision
Where the goal is to learn the relationship between training Machine Learning
inputs and training targets— Three specific types of What is Machine Learning
ML Workflow
machine-learning problems: 11 ML Types

) Binary classification, Classifying movie reviews as References

positive or negative
) multiclass classification, Classifying news wires by topic
) scalar regression, Estimating the price of a house, given
real-estate data

15
Unsupervised ML

Introduction
Where are we inCS?

Pattern recognition
Definition
Computer Vision

) We just provide ML with a large amount of data and Machine Learning


What is Machine Learning
characteristic of each observation (single piece of data). ML Workflow

12 ML Types
) Example: Images of cats and dogs without labels. References
However, you still want to split this data into 2 categories.
) You can employ unsupervised ML (in this case a technique
called clustering) to separate images in two groups based
on some inherent features(characteristics) of the pictures.

15
reinforcement learning

Introduction
Where are we inCS?

Pattern recognition
Definition
Computer Vision

Machine Learning
) In reinforcement learning, an agent receives information What is Machine Learning

about its environment and learns to choose actions that ML Workflow


13 ML Types
will maximize some reward. References
) Example, a neural network (model) that “looks” at a
videogame screen and outputs game actions in order to
maximize its score can be trained via reinforcement
learning.

15
References

ML Course

Introduction
Where are we inCS?

Pattern recognition
Definition
Computer Vision
) Online: Machine Learning

http://www.computervisionblog.com/2015/03/deep- What is Machine Learning


ML Workflow

learning-vs-machine-learning-vs.html ML Types

14 References
) Online: http://nkonst.com/machine-learning-explained-
simple-words/
) Book: François Chollet, Deep learning with Python

15

You might also like