Machine Learning
Machine Learning is the subfield of computer science that gives computer the ability to learn
without being explicitly programmed.
What does machine learning work?
It analyzes multiple data and predict object accordingly. Like if we give it a set of pictures of a
set of dogs and cats and ask it to differentiate between them then for a normal program we
have to give a lot of rules but in machine learning it analyzes the data such size of eyes and tail,
size of tail and etcetera and then decides accordingly.
Examples
Netflix and Amazon suggest you TV shows according to your previous choices of genres,
languages, etc. Banks approve loan by using machine learning as they search through the
probability of if the applicant can be given loan. Telecommunication companies use their
customers demographic data to segment them or predict that if they will unsubscribe them.
Differences between AI, machine and deep learning
AI components:
Computer vision
Language processing
Creativity
Summarization
Machine Learning:
Classification
Clustering
Neural Network
Revolution in ML:
Deep Learning
Machine Learning Categories
Machine Learning
Supervised Learning Unsupervised Learning
Regression – Predicting continuous values Clustering – Finding structure in the data
Classification – Predicting the category of an
observation
Branches of machine learning
Deep Learning – Deals with algorithm and human processing
Natural language processing – Deals with understanding how machines understand
human language.
Computer Vision – Deals with how a computer deals with a digital image
Reinforcement learning – Deals with teaching a machine to make decisions by
rewarding desired actions and punishing undesired actions.
Machine Learning History
In the machine learning history video you learned about some
important milestone in the history of machine learning. In this
section, we will provide you with some resources to explore some
of these events.
The Turing Test
The Turing Test was a test developed by Alan Turing to determine if
a machine can think like a human. According to the test, a machine
can think like a human if can fool a human into believing it is a
human when engaging in a conversation with another human. To
learn more about the Turing Test you can watch this video "What is
the Turing Test?".
Watson's Victory in Jeopardy
Ken Jennings and Brad Rutter were some of the best contestants on
the TV quiz show Jeopardy and in 2011 they agreed to play against
IBM watson.The IBM Watson supercomputer beat two of the most
successful human contestants in a game of Jeopardy. This victory
showcased a computers ability to search through vast amounts of
textual information, find potential answers and reason as to which
of the potential answers would be the right one. It also showcases
how computers that can question and respond can make its way to
our day to day lives. To watch how IBM watson beat humans in the
game of jeopardy and help understand what this victory means in
natural language understanding, you can watch the video Watson
and the Jeopardy! Challenge.
Project Debator
IBM developed Project Debator, an AI system that can take part in a
live debate with humans. Human language is complicated and
language is how humans express and formulate their thoughts. To
learn more about the project debator and watch it engage in a live
debate with Harish Natarajan, a grand finalist at the 2016 World
debating championships and a graduate of the University of Oxford
and the University of Cambridge, you can watch the video "IBM
Project Debater".
Codex
OpenAI's GPT-3 is a natural language model that can produce
human-like text with a given prompt. After GPT-3, OpenAI has
created codex which has similar natural understanding as the GPT-
3 but writes code given a prompt in the english language. To watch
a live demo of codex please click here.
