Scribd
Upload
2 views

NPTEL_Week01_02_OverviewOfMachineLearning.pptx

The document provides an overview of machine learning, defining key terms such as artificial intelligence, machine learning, and deep learning. It discusses when machine learning is useful, types of learning approaches, and outlines the seven steps involved in machine learning processes. Additionally, it highlights the mathematical concepts relevant to machine learning, including linear algebra, probability, and optimization.

Uploaded by

Ijaz Fazil Syed Ahmed Kabir
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
2 views

NPTEL_Week01_02_OverviewOfMachineLearning.pptx

The document provides an overview of machine learning, defining key terms such as artificial intelligence, machine learning, and deep learning. It discusses when machine learning is useful, types of learning approaches, and outlines the seven steps involved in machine learning processes. Additionally, it highlights the mathematical concepts relevant to machine learning, including linear algebra, probability, and optimization.

Uploaded by

Ijaz Fazil Syed Ahmed Kabir
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/ 12

Machine Learning for Engineering and

Science Applications

Overview of Machine Learning


Some common terms
■ Artificial Intelligence – Any method that tries to replicate
the results of some aspect of human cognition

■ Machine Learning – Programs that perform better with


experience.

■ Artificial Neural Networks (ANN) – A Machine Learning


algorithm

■ Deep Learning – A type of ANN

■ Big Data – Using data to find unobvious patterns


The AI Venn diagram

Machine
Learning

Deep Learning

Artificial Intelligence

Adapted from Deep Learning, Goodfellow et al (2016)


What is Machine Learning?

■ Simple Definition -- Using Data to answer questions


■ Study of computer algorithms
❑ that improve automatically
❑ through experience.
■ Formally, A computer program is said to learn from experience E with
respect to some class of tasks T and performance measure P if its
performance at tasks in T, as measured by P, improves with experience E.
https://upload.wikimedia.org/wikipedia/commons/thumb/6/6a/Maching_Reading_Robot_Auto-Text_to_Knowledge.jpg/463px-Maching_Reading_Robot_Auto-Text_to_Kno
wledge.jpg
Yufeng Guo -- https://www.youtube.com/watch?v=HcqpanDadyQ
https://upload.wikimedia.org/wikipedia/commons/thumb/7/72/SVM_margin.png/1280px-SVM_margin.png
The Machine Learning Paradigm
Rules

Classical Answers
Programming
Data

Data

Machine Rules
Learning
Answers

Idea adapted from Pedro Domingo (multiple sources) and Francois Chollet (Deep Learning with Python)
When is Machine Learning useful?
■ When experts are unable to explain their expertise
❑ Image recognition
❑ Speech recognition
❑ Driving a car

■ When Human expertise does not exist


❑ Hazardous environments -- Navigating on Mars

■ Solution needs to be adapted to particular cases


❑ User biometrics
❑ Patient specific treatments
A fundamental “trick” in most of ML
■ All problems are data, all solutions are functions/maps

■ Cognitive tasks -- Humans get sensory inputs as qualia


❑ We must convert these qualitative inputs into numbers – Input Vectors
❑ Similarly, outputs that humans give must also be converted into numbers
– Output/Target vectors

■ Determining appropriate inputs and outputs for a machine


learning task is an essential part of the process

■ Often the “Learning Task” is learning the mapping from input


to output.
Types of learning approaches
■ Supervised Learning
❑ Data labeled by human experts
❑ Labeling images
❑ Speech recognition
❑ OCR

■ Unsupervised Learning
❑ Unlabeled data
❑ Grouping customers
❑ Detecting new diseases
❑ Anomaly detection
Other types of learning approaches
■ Generative approaches
❑ Creating new data that is “like” given data
❑ Generally included in unsupervised learning

■ Semi-supervised learning
❑ Small amount of labeled data available along with unlabeled data

■ Self-supervised learning
❑ Implicit labels are extracted from data using heuristics

■ Reinforcement learning
❑ Actions are chosen based on rewards. Example : Chess, Games, etc
❑ Feedback is far removed temporally from action

The distinction between the various types of learning is often blurred


Seven Steps in Machine Learning
1. Gathering Data
❑ Deciding what “data” means is part of the problem
2. Preparing Data
❑ Ensuring that there is no bias
3. Choosing a Model/Algorithm
❑ Examples – Random Forest, ANNs, Hidden Markov Models,etc
4. Training
❑ Using data to determine model parameters
5. Evaluation – How well did we do?
6. Hyperparameter Tuning
7. Prediction

Yufeng Guo -- https://www.youtube.com/watch?v=nKW8Ndu7Mjw


Two problems in Supervised Learning

Classification Regression
Split it Fit it
Discrete or Categorical data. Real number data
Has category associated Has associated number
Example : Tumour classification Example : Prediction of stock market
Mathematical ideas we will be using in this course
■ Linear Algebra
❑ Machine Learning involves mapping
■ From vectors to vectors
■ That is, matrix based transformations – Linear Algebra

■ Probability
❑ Data and results have uncertainty built into them
❑ Conditional probability is a very important component of ML

■ Optimization
❑ For a given set of data what is the “best” model?
❑ Many ML models finally reduce to solving some optimization problem

You might also like