Scribd
Upload
2K views

Hands-On Machine Learning With Scikit-Learn and TensorFlow

The document provides an overview of machine learning concepts, including: - Supervised learning involves training data that includes labels, and is used for classification and regression. Unsupervised learning uses unlabeled training data for tasks like clustering and dimensionality reduction. - Machine learning can simplify programming complex tasks, perform better than rule-based systems, and provide insights from large datasets. Popular supervised algorithms include decision trees, neural networks, and support vector machines. - Reinforcement learning involves an agent that interacts with an environment and receives rewards or penalties to learn optimal behaviors or policies through trial and error. Batch learning uses all data at once while online learning allows incremental learning from streaming data.

Uploaded by

Hiba Khurshid
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
2K views

Hands-On Machine Learning With Scikit-Learn and TensorFlow

The document provides an overview of machine learning concepts, including: - Supervised learning involves training data that includes labels, and is used for classification and regression. Unsupervised learning uses unlabeled training data for tasks like clustering and dimensionality reduction. - Machine learning can simplify programming complex tasks, perform better than rule-based systems, and provide insights from large datasets. Popular supervised algorithms include decision trees, neural networks, and support vector machines. - Reinforcement learning involves an agent that interacts with an environment and receives rewards or penalties to learn optimal behaviors or policies through trial and error. Batch learning uses all data at once while online learning allows incremental learning from streaming data.

Uploaded by

Hiba Khurshid
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 18

Hands-on Machine Learning

with Scikit-Learn and


TensorFlow
CONCEPTS, TOOLS, AND TECHNIQUES TO BUID INTELLIGENT SYSTEMS
History:

 Machine Learning has been around for decades.


 Optical Character Recognition (OCR).
 Spam Filter-First application (1990s)
Road Map

 Supervised v/s unsupervised learning


 Online v/s batch learning
 Instance based v/s model based learning.
 Work flow of the typical ML project.
 Main challenges you may face
 Evaluate and fine-tune a ML system.

Book link:
What is Machine Learning:

 Science of programming computers so they can learn from data


 Field of study that gives computes the ability to learn without being explicitly
programmed. [Arthur Samuel, 1959]
Engineering-oriented one:

A computer program is said to learn from experience E, with respect to some task T and some
performance measure P, if its performance on T, as measured by P, improves with experience E.

[Tom Mitchell,1997]
 Training set: Examples that are use to learn.
 Training instance: Each training example.
 Accuracy: performance measure.
 Testing set: ?
Why use Machine Learning:

 Traditional programming will likely become a long list of complex rules.


 Pretty hard to maintain.
 Automatically learns.
 Program is much shorter, easier to maintain and most likely more accurate.
Traditional Programming v/s ML
Machine Learning is great for :

 To simplify code and perform better.


 Complex problems where there is no good solution.
 Fluctuating environment.
 Getting insight about complex problems and large amount of data.
Types of Machine Learning Systems

 Machine learning systems can be categorized based on :

 Trained with human supervision.


 They can learn incrementally on the fly.
 Detect pattern in the training data and build a predictive model.
Supervised/Unsupervised Learning:

 Machine Learning systems can be classified according to the amount and type of
supervision.

 Supervised Learning
 Unsupervised Learning
 Semi supervised Learning
 Reinforcement Learning
Supervised Learning

 The training data you feed to the algorithm includes the desired solutions, called labels
 Supervised learning tasks:

 Classification (Spam filter)

 Regression – to predict the target numeric value (price of house )

 Logistic Regression – used for classification


 Output a value that corresponds to the probability of belonging class.
 Eg. 20% chance of being spam
Supervised Learning Algorithms:

 K-Nearest Neighbors
 Linear Regression
 Logistic Regression
 Support Vector machines (SVMs)
 Decision Trees and Random Forests.
 Neural Networks (can be unsupervised or semisupervised)
Unsupervised Learning:

 The training data is un labeled


 System tries to learn without a teacher
Unsupervised Learning Algorithms

 Clustering :
 K-means
 Hierarchical Cluster Analysis(HCA)
 Expectation Maximization
 Visualization and dimensionality reduction
 Principal Component Analysis
 Kernel PCA
 Locally –Linear Embedding(LLE)
 Associate rule Learning…….?
 Apriori
 Eclat
Semi supervised Learning

 Lots of unlabeled data and a little bit of labeled data


 Most algorithms are combination of unsupervised and supervised algorithms
 Google photo
 Person A shows up in photo 1,2,5 and 7 (unsupervised)
 Name of that person (supervised)
Reinforcement Learning:

 The learning system , called an agent in this context, can observe the environment ,
select and performs actions.
 Get rewards in return or penalties in the form of negative rewards.
 It learns by itself the best strategy, called policy.
 Example Robot
 Deep Mind’s AlphaGo Program
 Beats the world champion Lee Sedol at the game of Go
Batch and Online Learning

 Whether or not a system can learn incrementally from a stream of incoming data
 Batch :
 System is incapable of learning incrementally.
 Take a lot of time and computing resources. So it is typically done offline

You might also like