Machine Learning for Everyone: Lattice Nepal

Machine Learning for Everyone: Lattice Nepal 1

INTRODUCTION

Machine Learning

Source: DATA-Flair.training

The term Machine Learning was first introduced by an American Arthur Samuel working in the field of computer gaming and AI in 1959, and quantified that “It gives computer the ability to learn without being explicitly programmed”. Machine learning is a subfield of artificial intelligence (AI). The objective of machine learning is to understand the structure of the data and fit that data into models that can be understood and utilized by people. Machines takes data and learn for themselves like human is the central principle of Machine Learning. The basic
principle of machine learning is to make algorithms that can receive input data and use statistical evaluation to forecast and output while updating outputs as new data becomes available. Some of the programming languages to write algorithms in ML are python, Lisp, R.R, java etc. The way of writing algorithms in machine learning can be divided into 3 ways and are supervised, unsupervised and reinforcement learning.

Current scenario.

The use of Machine learning in our practical life has been increasing and showing no sign to stop at all. Most of the Researchers in the world think that Machine learning is the best way to make progress toward human-level AI. Any technological user today has been benefitted from the uses of Machine learning. Virtual Personal Assistant is an outcome of the use of Machine Learning and is currently used by leading tech giants like google, Amazon, Apple, Samsung etc. on their devices. Machine Learning is used in filtering of spammed Emails and malware filtering and
over 0.3 million malwares are detected. YouTube and many video contents providing applications like Netflix, Amazon primes etc. uses recommendation  engines which are powered by Machine Learning. Others examples of use of ML are Facial Recognition technology, optical character readers, self-driving cars and many
mores. According to the popular business magazine called Forbes the business values created by AI and ML will reach 3.9 trillion US dollar in the end of 2020. The report also said only 18% of business firm adopting machine learning are said to have not gotten financial return.

BACKGROUND

TYPES OF MACHINE LEARNING

Simply, allowing computers to learn by designing different algorithm is Machine Learning. There are many ways of writing machine learning algorithms like supervised, unsupervised, semi-supervised, reinforcement, transduction and learning to learn but mostly commonly used are described below:

SUPERVISED LEARNING

The process of gaining information, statistics and facts with the help of data provided can be considered as supervised learning. We provide a set of data which acts as a guide and automatically helps the machine to train. once the data are fed up to the computer it starts making predictions and decisions. In this type of learning we
already know what our expected output should look like. The two types of  supervised learning are Regression and classification. The examples of supervised learnings are Weather forecasting, population growth prediction, image
classification and so on.

UNSUPERVISED LEARNING

In Unsupervised Learning, the data are provided to the computers and goal is to find the relationship and patterns in the dataset by creating clusters in it. In this type of learning we simply try to find the resemblances in training data. we don’t have any idea what our final output will look like in this of learnings. Clustering and dimensionally reduction method are two ways of writing unsupervised methods of algorithms. This way of writing algorithms is difficult than supervised learning and has produced many success like machine capable of self-driving, Recommending system just like in YouTube, Netflix, people you may know options in Facebook, targeted marketing, big data visualization and so on.

REINFORCEMENT LEARNING

Reinforcement learning is one of the three basics machine learning paradigms. Reinforcement learning can be considered as a dynamic learning that trains algorithms using different ways of Reward and Punishment. In even simpler terms, it gains data by interacting with the environment. The theory of reinforcement learning depends upon dynamic programming and artificial intelligence. Reinforcement learning makes the decision chronologically. The key feature of this type of learning is gains new behavior and skills incrementally. Reinforcement learning is used in various tasks like robot navigation, gaming AI, real time decisions, task acquisition, learning tools and so on.

Its been while, Nepal entered the age of digitalization but we have witnessed the technological Revolution so far so good. There were many innovative startups as some creative and awesome tech companies like fuse machine, deer walk Inc. leapfrog technology, paaila technology, Agile IT solutions, Beery bytes and so on has taken responsibility in bringing positive change in the country and society. Nepalese machine learning enthusiasts are optimistic as some Nepalese based companies like fuse machine and others are providing scholarships, interactive sessions to develop talent in the country. The scope of machine learning in Nepal is growing day by day and the best example of it is Nepal’s biggest online selling sites like daraz and sasto deal using it to provide personalized experience to show better products as per the requirement of clients. Another example of use of machine learning in Nepal is PARI 2.0 located at the Nepal SBI bank which recognizes the customer and greet and was developed by paaila technology. The revenues generated in Nepal are unknown but the demand of Machine learning programmers is increasing day by day and the market of machine learning based product is brilliant.

World is facing many challenges related to different field and Machine learning has tackled some of the real-world challenges in best possible ways. Machine learning can be implemented in many sectors like communication, health, technology related fields, government offices but the use of it in our Nepal has been very limited due to various reasons. The first and foremost use of machine learning in our country must be done in government offices and police office as the process are tedious, boring and very time consuming. Our economy is based on agriculture and agriculture sector can grow at a significant rate if our farmers tends to use technologies aided through machine learning. The technology can be used in detecting the nature of soil by supplying the necessary data. furthermore, they can be used to analyze the amount of water and fertilizers needed for crop. With the help of machine learning aided robots, drones we can spray the required amount of pesticides, insecticides, and water. Machine learning algorithms are used to predict natural calamities like weather forecasting, landslides on hilly areas, wildfires, and so on. We can use machine learning robots to fight natural calamities like wildfire. It can predict the
direction and behavior of wildfires very effectively and pass the data to relevant management offices. The use of human manpower to stop such huge fire can be very dreadful. So, by using different patterns of machine learning in robots we can rescue animals, birds and act on stoppage of wildfire. Moreover, through ML aided drones we can provide required amount of foods and other stuffs to animals.

It is a bit difficult to implement the use of machine learning. If the use of machine learning is to be done, there must be plans and ideas which must be followed. Machine learning can be used in good governance. Machine learning allows government to deliver better, cost effective and citizen friendly services. The first and important plan is to invest good money in building manpower which can use machine learning algorithms. Civil servants and police officers can be assigned more effectively in required and crowded areas if more computational power, larger data
sets and improvements to algorithms are done. Using social media monitoring, computer vision and data annotation government agencies can extract the required information when suspected and monitor on national threat. Emergencies like coordinated attacks and calamities can occur at any moment. Life’s are at stake at
interaction is much. So, by implementing the machine learning application like translation, voice recognition, text data collection, and weather forecasting responders can interact with officials and people and save themselves.

Conclusion

The world and the future of human being will be better due to the advancement of machine learning. more Many scientists believe machine learning is the best and only one way to reach towards human level AI. As everything in the world has advantages and disadvantage so does machine learning have but the advantages of it outcast its disadvantages. A lot of progress has to be done in algorithms related to reinforcement learning to reach towards human level artificial intelligence. Weather  forecasting, population growth prediction, image classification has brought pretty good changes in the way government operates. Machine learning has revolutionized the world and how we operate to some extent.

Future escalation.

Self-driving cars, designing a drug to target a specific disease, medical record mining and many more are believed to be done using machine learning in future. experts believe the ratio of jobs created and hampered due to the advancement of machine learning in present and future will be equal. Let’s hope the advancement in AI and
ML will not outcast the performance of human.

 

To Know about 3d Printer: https://latticenepal.com/3d-printer-advantages/

Leave a Reply

Your email address will not be published. Required fields are marked *