About Machine Learning System (M-LMS)

About Machine Learning System (M-LMS)


What is machine learning?

Every person-object-animal scattered around you is following a certain time-to-time pattern. Combining information from the past history of something enriches our experience when you carefully consider what might happen to that thing in the future. Now if we can use this same thing machine/computer to extract a pattern of something with enough past history/data and from there we try to know a possible future outcome this whole thing is called machine learning. Simply put, you feed the machine various historical/past/experienced data -> the machine analyzes it and builds models/finds patterns -> finally predicts the future results from it, that’s machine learning.

Usually through programming we give detailed instructions to the computer to work. But that is not done in the field of machine learning. In this case, the data is given to the machine learning model. The machine learning model finds and classifies different patterns in the data. Through which the computer can predict/recognize different things.


Types of Machine Learning:

Machine learning is mainly of 2 types.

  • 1) Supervised learning.
  • 2) Unsupervised learning.

Supervised Learning:

In this case, the machine learning model is trained with labeled data. The data given to the model contains various features and labels. Now an example can be given to understand features and labels. Suppose a child is given an orange and an apple. He is shown the properties of orange and apple and told that this is an orange and this is an apple. Later he can easily recognize which is an orange and which is an apple. In this case orange is a label and features of orange are features. Again apple is a label and features of apple are features.

The data used in supervised learning consists of labels and its features. As in the above example, the computer is first recognized/taught. Later, it can recognize itself. That is, it can predict labels based on features. Linear regression, random forest, Support Vector Machine Supervised Machine Learning Algorithms.

Unsupervised Learning:

In this case, the machine learning model is not trained with labels and its features. The model classifies the data based on the similarity between the data. Suppose a child who does not know oranges and apples is given 10 oranges and 10 apples in a basket. He doesn’t know which is an orange and which is an apple. Now he is asked to put those that are similar together. He must be able to divide the fruits into two, one part of which will be orange (10) and the other part will be apple (10). The child could do the work without knowing whether it is an orange or an apple. Because an orange and an apple have no similarity.

Again, there is a complete similarity between two oranges. He is able to divide them into 2 by seeing this similarity and dissimilarity. In case of unsupervised learning, MEDAL works like this. The model separates or classifies the data based on various similarities and differences. Unsupervised Machine Learning Algorithms like k-means clustering, KNN etc.

Machine Learning and Deep Learning:

Artificial intelligence is a subset of machine learning and deep learning is a subset of machine learning. So we can call deep learning as machine learning. We can thus say that deep learning is unsupervised machine learning where unstructured data is used. The key here is unstructured data. When we take structured data it will be supervised/unsupervised machine learning.

But when we take unstructured data, it will become deep learning. In supervised / unsupervised machine learning, raw data is processed into structured data and the model is trained with that structured data. And in deep learning, raw data is directly input. Eg. Neural Networks (Deep Learning) find and classify various features in that data. An example can be given.

First let’s say what is features extraction. Suppose a model is given images of fruits such as mango, jam, jackfruit etc. as data and their characteristics are given. In this case the label is mango, jam, jackfruit etc. and its features are features. Here we tell a model that the characteristics of mango, jam, jackfruit etc. These are This work is called features extraction. In general machine learning, features are extracted from the data and fed to the model.

But in the case of deep learning, the programmer does not do the job of expression of these features. The programmer creates the neural network which does the job of feature extraction and classification by itself. In this case, the issue is that in the neural network, we give as data the image of mango, it is a picture of a tree etc. But we do not say what will be the features of the mango/jam/tree. The neural network automatically understands what the mango/jam/jam features are and classifies them accordingly.

Now the question may come why we should use deep learning. What benefits will we get as a result of deep learning?

Facebook stores the data of a Facebook user. What kind of posts he likes. What kind of posts he comments on. What kind of groups he is active in. His geolocation. What kind of people he likes to talk to. And various other data. The data is huge and Different types. In this case, the work of features extraction becomes difficult. In this case, deep learning technique is useful. Again, Google or other search engines also have different types and huge amount of data. In this case, it is never possible to do features extraction.

Deep learning is very useful when working with large amount of data. Because in that case feature extraction/structuring the data is difficult or impossible. In this case we use deep learning. Deep learning is used in voice recognition, language translation, object detection, search engines etc.

All you need to know to learn machine learning:

Must have good understanding of Python/R (R) programming language. Must know about data structures and algorithms. To work with machine learning you need to know the following libraries of Python programming language:

  • 1) Nampai.
  • 2) Pandas.
  • 3) Metplotlib.
  • 4) Seaborne.
  • 5) Psychit-Learn.
  • 6) Saipai.
  • 7) Tensorflow

Also, to work with machine learning, you need to know linear algebra, various formulas of statistics, calculus, probability theory.

Applications of Machine Learning:

Based on user data on YouTube/Facebook/Netflix, the machine learning model predicts what type of content the user likes. The type of content that the user likes to watch is suggested to him through machine learning. Besides, machine learning is used to predict the weather of different regions. Speech recognition, object detection is done through machine learning. Machine learning is used in self-driving cars so that the car can run on its own. Machine learning is also used in many other sectors.

What is a pipeline in machine learning?

It is important to build any machine learning model and have enough data to train the model. When a model is deployed in real life, these data are collected from different places. These collected data may be in various formats. This is why data cleaning and processing is seen as an important step in machine learning. New data collection is an ongoing process in machine learning. Whenever new data is added to the existing dataset, the same data cleaning and processing steps have to be repeated on the newly added data. It is a tedious, tedious and time-consuming process.

An alternative to this monotonous, tedious and time-consuming process is to build a machine learning pipeline. Machine learning pipeline – remembers the complete data cleaning and processing steps performed previously and whenever new data arrives, performs predefined data cleaning and processing steps on the new data. This makes the machine learning model very easily updated and efficient.

Machine learning and AI

Its is a technique of computer programming in which a computer program is designed in such a way that the program can learn new things on its own and make decisions on its own when needed. This technique in computer field is called machine learning. Because machine learning is machine learning itself. So any application or software is designed in such a way that it can learn new things without any intervention in its program and predict or output information related to that data when given time.

Artificial intelligence is abbreviated AI whose full form is Artificial Intelligence, which means artificial intelligence or artificial brain in Bengali. This scientific technology imparts human intelligence to machines made of metal. And their brains are so developed that they are able to think and act like humans.

This technology is mainly used on computer systems to do this mainly three processes have to go through. Which is – firstly learning which means the machines are taught what work they have to do and what rules to follow to do those works, secondly Reasoning i.e. this is what they teach the machines to move towards the right result. Rules were created to instruct them to follow so that they can reach a certain result. Thirdly, Self-Correction is the process by which machines are taught how to correct mistakes.

Essentials of learning machine learning

When you hear the name machine learning, you should be able to guess what it really is. To be able to operate a machine i.e. a computer/device in such a way that it can function much like an intelligent human being. Not that everyone should know about it. But it is true that this method will be useful in your daily life somehow.

Even when you are searching for something on the internet, you are using machine learning without even realizing it. But this machine learning thing is seen very differently in our country. Many people think that machine learning is not just about importing some functions from some libraries, but a big part of it is mathematics and statistics. If you really want to know machine learning then you need to know about these. Otherwise, machine learning will seem like a black box to you. You can read this post to know what is there in machine learning or data science.

Machine learning is one of the trending topics of the present time. Currently, there is an effort to make everything autonomous in the world. And machine learning is playing one of the roles behind this autonomy. Countless problems that humans never thought a machine could do are now being done with machine learning. A slightly more advanced method than machine learning is now called deep learning. If you think that you want to make a technological contribution to the world today, machine learning is a very good thing, there are many tutorials and books on machine learning that will help you learn.

Checkout this article: Academic Learning Management System

Sign up to receive awesome content in your inbox, every month.

Leave a Comment

Scroll to Top