Supervised Machine Learning

Vinoth Rathinam
A free video tutorial from Vinoth Rathinam
Data Scientist | Automation Architect | Technical Trainer
4.3 instructor rating • 3 courses • 78,715 students

Lecture description

This video explains the below topics : 

Definition of ML

How Supervised Learning works? 

Definition of Supervised learning

Difference between Classification vs Regression Model

List of Classification Algorithms

List of Regression Algorithms

Sample output of Classification and Regression Model

How to choose Algorithm for your Project?

Learn more from the full course

Welcome to Artificial Intelligence !

NON TECHNICAL COURSE specifically created for AI/ML/DL Aspirants, gives insight about Road map to A.I

48:42 of on-demand video • Updated January 2019

  • Basic Idea of Artificial Intelligence and Machine Learning
  • Prerequisites or Road map to start Machine learning project(ML)
  • How to choose the best programming language for AI ?
  • How much Mathematical knowledge needed for AI ?
  • Which is the best AI Engine/Tool/Framework for AI ?
  • Why do we need to learn Algorithm?
  • Types of Machine Learning Algorithms with Real time scenario examples
English [Auto] Hello everyone. Welcome to Reno tracking number. Ideals. There are three types of machine learning. Supervised Learning and supervised learning and reinforcement learning. In this video we are going to learn about supervised learning with simple example. A short recap definition of mission learning mission learning is the subset of artificial intelligence it means in artificial intelligence we have many ideas in that machine learning is one of the subset. So we should not think artificial intelligence on machine learning. I've seen both are different then which works based on self-learning algorithm using past experience our dataset. We thought being explicitly programmed it means the self learning algorithm even works for our new scenario which is not programmed just to make this concept easy. Understanding and fun learning. I took this example a very famous fictional superhero movie Iron Man that Tony Stark is a hero and he will be having a powerful suit of armor this armor sword guides him to identify the vehicle building missiles enemies target and so on. But in this example he won't do brain the algorithm and makes this bar full suit to identify whether the images dog or not dog here the human teaching the missions. That's why it does guard supervised learning. I'm saying it again the human is teaching the mission is called a supervised learning. Now we will see how it works. He had that Tony Foster uses that training data set which consist of different types of dog images. The key PTSD. We have our own three hundred and thirty nine dog breeds in the world. Then the training data set should contain multiple images of all the three hundred and thirty nine dogs and based on parameters like breed color size eyes nose dung and whether it's poppy or not. I over that the dog is dirt. They are not like that. Now you can imagine how big the training dataset will be. Finally the algorithm learns that input back then and it generates the expected output. The algorithm will label eat image as dog and not dog. This process is keep on repeating until we get the final trained model. Sometimes we call it test predictive model. Now Tony want to test this predictive model with new to this data. First Dog in dress holding a cat and a cat in dress. They're trained to model are the trained algorithm are the predictable than can easily predict the output of that never seen input are unlabeled image. No the sword will exactly tell whether it's a dog or not. Hope you understood how supervised a mission learning works in chart. A supervisor uses training data set to create and train the predictive model then uses the unseen data are this data set to find the output of unlabeled input. No. Tell me which is easier to identify whether a MSA ordered Doc the hand when the number of parameter increases the complexity of the model will increase. So just think about it. Book definition supervised learning is a mission learning task of determining a function from labeled data. In our example the machine learning algorithm that detects if the images dog or not the training data set would include the image labeled as dog and not a dog to help each child go to them how to recognize the difference. So the supervised learning algorithm interferes of functions from a label to data set and used as functions on new examples are never seen this data. Hope you got some idea bought out supervised learning to tapes of outputs produced by the supervised learning it will be a discrete value and continuous value which comes under the classification and regression problems classification. I'll go to them as used when they decide our purpose at discrete label. Nothing but a binary classification where only two possible outcomes like s are all here. Two examples I am mentioned. Just to identify whether the e-mail is spam or not. Then that transaction is fraud. Are straw transaction that's about the classification problems. Then second on this regression regression algorithms are used when the desired output is that you are continuous value. It means the answer to your question is that represented by your quantity. For example predicting the house price based on area then predicting the number of copies of music or album will be sold in next month. These examples falls under the regression model can see some list of classification algorithm like logistic regression decision tree kernel approximation gain nearest neighbors random forest. These are all very important algorithm which is used for the classification models. So again these are the list of regression model algorithm like ordinary least Esq. regression linear regression then random photos also used in the regression model. Now a big ghosting comes to you. How should I choose the algorithm. It's the most confusing question for most of the data scientist first. Once you choose your project first identify whether it flies under the supervised learning or not. If it falls under the supervised learning then you how to divide it into two categories but it falls under the classification or it falls under the regression. In case if it falls under the classification then you have to choose which best algorithm will choose for a classification problem because some pains for your particular project. You can use maybe more than 2 to 3 algorithms but how to choose feasible solution. Same way if it falls under the regression then you how to choose D proper regression algorithm. We have completed theoretically the concept of supervised learning and also understood the difference between the classification and regression model. Surely you'll be learning how to create these types of graph practically using the linear algebra and all the important supervised machine learning algorithms in the upcoming videos that stand off today's session. Thanks for watching. Happy learning.