In machine learning and AI, you have to focus on a large scale of programming and mathematics to proficient in those two. Machine learning is a subset of AI and it is the path to learning AI. So you must have a lot of experience in machine learning to process with AI. Experience can be gain by doing more and more projects by yourself and learning machine learning is something that we can you with. So let’s get an idea of ‘what is machine learning and what is AI’.
In machine learning and AI, you have to focus on a large scale of programming and mathematics to proficient in those two. Machine learning is a subset of AI and it is the path to learning AI.
What is Machine Learning?
We know that humans learn from past experiences and machines follow instructions given by their owners. If humans can train the machines to learn from past data or from information to do the work it will be fast and that’s called machine learning. As an example let’s take you like watching movies, further you love to watch more violent and more romantic movies. Let’s represent that on a graph.
In here suppose that you dislike less violence and less romantic movies. Let’s think of a new movie with medium violence and medium Romantic, in this situation we cannot decide you dislike the movie or not. In machine learning, it will recollect your past date and put it in the majority votes side.
Machine learning will predict and give the best solution based on past data. If there are more data and a better model, accuracy will increase.
There are mainly 3 types in which machine learns
- Supervised learning
- Unsupervised learning
In supervised learning, the machine will study the past data to present data using labeled examples and predict the future. As an example let’s take three coins Rs.2, Rs.5 and Rs.10 and suppose their weights 1grams, 3grams and 6grams respectively. If we feed these data to the machine, weights as the feature and coin value as the label, the machine will label the feature with the corresponding coin value. So if you give a certain weight as input, the machine will study the features and output the coin value. This is called supervised learning.
In unsupervised learning, if you give a set of data to the machine it will divide the data into separate groups. But it will not depend on any labels. Unsupervised learning differs from supervised learning because here machine will not use labels to predict.
Reinforced learning based on the feedback that you will provide. As an example, Let’s think that you give a picture of a mouse and give instructions to the machine to identify it. Suppose that the machine will output it as a cat’s image, and that’s a wrong output. So you give negative feedback to the machine and say it’s a picture of the mouse. The machine will learn from that and when you give a picture of the dog it will identify it correctly. This learning method uses the trial and error method.
What is AI ?
Let’s picture this, a machine makes customized coffees for your family members or reorganizes your bookshelf as you like it. These are the product of artificial intelligence. This artificial intelligence builds with complex algorithms and mathematical functions. The AI is used in smartphones, automatic car systems, banking, video games and in many other aspects of our day to day life. Generalized learning, reasoning and problem solving make a machine artificially intelligent. AI provides machines the capabilities to adapt, reason and provide solutions. AI categorized into two parts
- Narrow AI
- Strong AI
Narrow AI focuses only on one task. It can be programed to do only one task. As an example, if a machine is programed to play checkers it cannot play chess instead of that. Strong AI can do multiple tasks like robots. But still strong AIs’ only appear in fiction.
To archive AI you need the machine learning and deep learning also.
Now let’s move to how to teach machine learning and AI
How to teach machine learning and AI?
First of all, if you are planning to teach machine learning and AI for students you have to divide them into two groups. The first group is who has no knowledge of programming and advance mathematics that means kids and the other group is those who have knowledge of advanced mathematics and programming to some extent.
How to teach machine learning and AI?
Machine learning and AI both are advanced fields for kids. So you need to give a good impression of what is machine learning and AI. Otherwise, they will get bored in the first place. You have to set their mind for learning machine learning by using examples like saying you can make robots and games. Then they will get interested in learning those things. Most kids don’t like to read books and they will never read an article full of codes. So we need to find another way to improvise kids to familiar with machine learning. The best solution to teach machine learning and AI is SCRATCH. Scratch is a block-based visual programming language specially creates for primary students. This is can help primary students to learn coding, machine learning and AI. SCRATCH got an attractive GUI (Graphical user interface) and it’s easy to use even for kids. In this platform, kids can give instructions and data to a machine to learn. The best thing here is kids don’t want to learn any codes to do these things. They only need text to give data to the machine. This will give an interesting feeling to a kid. After a few demonstrations, kids can create their own projects and it will develop the ability to think of new creations.
As an example, draw a fan and a light on the Scratch drawing area. You want to turn on the fan when you say turn on or when you say it’s hot in here. Scratch will visually turn the fan on. So like that let’s suppose you need to turn on the fan, turn off the fan, turn on the light and turn off the light. In scratch, it will provide you to draw 4 tables under the above topics and you have to fill the table with your own command texts, like turn on, it’s hot in here, need cool, need light, so dark. Kids can fill these tables with their own imaginations and try a test run on the scratch virtual interface. You can give this as group work for kids so they can learn machine learning and AI at the same time they can get other ideas and improve their own imagination. This will help you to teach machine learning and AI to kids in a fun way. The main advantage of scratch is kids can do projects without using a single code. So it is easy and meaningful than teaching codes at once. Kids also get a chance to improve their own thinking ability by planning different projects in Scratch.
When kids are comfortable with scratch they will have an idea about what they are doing and what machine learning and AI looks like. Then you have to introduce them to a text-based language like LOGO which is based on both graphical and texture platforms. LOGO gives its users, facility to learn programming by just using words and directions. Users don’t need to memorize or use any complicated programming structures. Because of this facility kids can easily be trained in programming using simple words. With their experience with scratch they can easily catch up on the methods in logo and it will help to improve their knowledge to the next level.
How to teach machine learning from beginner level to advance level
Above we talk about how to teach machine learning and AI in kid’s version, now in here you can get the knowledge on how to teach machine learning and AI in an advanced way. When talking about advanced machine learning it includes several advanced mathematical theories. Probability, Statistics and linear algebra are used in machine learning in order to get a better solution. These mathematical theories are not necessary for beginner level, but when it comes to the advanced level some theories of the above fields will be used in some projects. As for the beginners you only need to give the knowledge of a programming language that will help them to create simple AI projects.
Machine learning mainly based on programming languages and advanced mathematics. A Programming language is a tool that we use to give instruction to a computer. Normally a computer, a phone or any computer base device uses 0’s and 1’s to calculate and perform the output. They use millions of 0’s and 1’s to comply and humans cannot understand these binary representations on large scale. Programming languages translate these 0’s and 1’s into a human readable format. Also, we can give instructions to the computer through a programming language. Mainly there are two types of languages
- Low level language
- High level language
Before teaching python you need to install an editor or you can execute the program in command promote by running the python file(.py). You can start teaching python by using simple code lines and give them a proper understanding of codes. As an example
print (“Hello World!”)
By running the above code it will print it as
Hereby using this code segment you can teach how punctuation working in python. If one bracket or punctuation missing, the code will not work properly. You can give them a chance to misplace punctuation and try to run and understand on their own how it’s important.
A beginner in python must have the knowledge of Python operators, arrays, conditional loops, Strings and functions. You need to focus on each and every above section to teach python properly. In python operators, you need to teach arithmetic, logic and comparison. In arrays, need to focus on how to create an array, pop and reverse an array with examples. In conditional loops the most important parts are IF, ELSE, ELIF, While loop and For loop. In Strings and Functions, there are several things that you need to teach beginners. After these sections anyone can read and understand other methods easily. You only need to give the basic knowledge of python. Because it is a large area, they need to study by themselves after having the basic knowledge.
You must also have knowledge of machine learning algorithms in order to learn AI and machine learning. There are three types of algorithms
- Linear Algorithms
- Gradient Descent
- Linear Regression
- Logistic Regression
- Linear Discriminant Analysis
- Nonlinear Algorithms
- Naïve Bayes
- K-nearest Neighbours
- Learning Vector Quantization
- Ensemble Algorithms
- Bagging and Random forest
- Boosting and AdaBoost
To study these algorithms you must have the knowledge of probability statistics and linear algebra. These mathematical methods are not compulsory for simple AI projects. But if you are planning to do a research study and create an AI system you must have the knowledge to understand these algorithms. To study algorithms you will need the below mathematical requirements.
- Probability foundational methods like Joint probability, marginal probability and conditional probability.
- Bayes Theorem- a principal way of calculating a conditional probability
- Probability distributions – give the possible values and their probability for a random variable.
- Statistical Hypothesis Tests
- Resampling methods
- Estimation statistics
- Linear Algebra
- N- Dimensional Arrays
- Matrix Factorization