Common algorithms

And here, I try to spend another session to explain a little bit about some simple machine learning algorithms, some common algorithms. As I already saw in Weka that you can perform like kNN random forest, so what are the idea behind their algorithms. And the first one is the K Nearest Neighbors algorithms, (KNN) KNN is a non-parametric method used for classification and regression. And as you can see in this picture, you can see that. you have two classes, the first one is the blue one, and the second one is the red one. If you have another data point in green color, and you want to classify whether the green color belongs to the blue color one or the red color one.

So you want to perform the kNN and the kNN idea is depends on the the distance algorithm. So they try to compare between the distance between the green one and blue one and red one. So if the distance is very nearby. It means that it is close to the class red. So it belongs to the class red and distant be close to the class blue. So it belongs to the class blue. And you can also select the number of case means that if the k closes in both… if the number of case closes with the new data points so you can treat the new data points as that classes. And the second one is the random forest.

And this is an ensemble learning and they combine the results between different decision trees, and then decide the results of the final tree. So which for this one we call this is the random forest. For example, here, you have a lot of trees here. And then for each tree will have a result and you combine all the results together. You can get another result and that is the example result for random forest. And the next one is for support vector machine. And a supervisor machine is like this figure. You can try to find the algorithms, the hyperplane that can help you to separate between two classes. And for the first figure, you can see that you can block..uh…

you can use alike to represent two classes. And if you only use the alike so you can separate two classes very easy. However, in the second figure, you can see that. if the data representation just like this, so you cannot use a light tools separate between two classes. In this case, you need to use a circle. Like I block in the green color to separate between two classes. And the difference with some traditional machine learning algorithms, you can… nowadays deep learning very common now. Because they can use uh… they can take advantage of a lot of data. And what is the difference here?

In the deep learning, it is an AI function that imitates the workings of human brains in processing data and creating patterns for use in decision making. And here is a subset of machine learning and also the AI, and here, this figure to show the process of deep learning. You can see that. In deep learning, there are a lot of layers inside. And it’s also known as the deep neural networks or also deep neural learning. And what is the difference between the shallow and also deep neural networks here? And just… different in on the hidden layers. Because you know that, for neural networks, there are three layers. The input layer hidden layers and also the output layer.

And there are two figures to show you the difference between the shallow and deep neural network just only difference in hidden layers. In the shallow, you have only one hidden layer. However, in deep neural networks you can even create a lot of layers inside the DNN. And for with a lot of layers, it can help you to learn the data very well and generate a good result, if you have a very big number in your data.