HSRW EOLab Wiki

Adults enjoy using a lot of sophisticated phrases to express themselves. This might be due to their desire to demonstrate their outstanding vocabulary. It's much worse in technical disciplines with a lot of huge technical terms, such as Computer Science. These technical terms are intended to assist distinguish between related concepts. However, it frequently confuses basic concepts. I'll try not to use any of these difficult terms in this essay. This is so that you can comprehend the concepts rather than memorizing large words. Let's begin with some background information.

Let’s take it from the top.
You may have seen computers all over the place. Computers come in a variety of shapes and sizes, from smartwatches to phones, tablets, and laptops. Today's handy devices have traditionally been referred to as “Dumb Machines.” They are as obedient as a dog when it comes to obeying directions (or kitten for those who are weird like that ). Swipe Right is what it does when you instruct it to. Swipe Left is what it does when you instruct it to. It will launch Instagram if you instruct it to when someone clicks on the Instagram symbol. I think you get the idea. It is incapable of thinking of anything on its own. Despite the fact that these machines are dumb, we have taught them to perform a lot of things during the previous 50 years. From adding numbers to performing tunes to defeating Grandmasters at Chess, there's something for everyone. But someone has to teach the computer what to do and how to do it for each of those things.As a result, the capabilities of a computer were restricted. Computers could only do tasks that humans could demonstrate to them (step by step). As you can expect, the enchantment quickly lost its allure. There are always those situations in which :

• Unordered List Item Even humans have no idea how to achieve it
• Ordered List Item certain duties for which we have no idea how to instruct the machine. (We can't express it well enough for a machine to grasp)

Computers couldn't accomplish anything without someone instructing them how to do it for a long time. Until recently, this was true. Some extremely clever people* devised a method for teaching computers to learn on their own. (Another popular term used by grownups is Machine Learning, but who needs that?) Consider being in a classroom with no teacher. You and your students study on your own without the help of an instructor. Isn't it strange? Imagine the strangeness of a computer learning on its own. Surprisingly, the computer is quite capable of self-learning. It can even learn complex things that humans are unable to teach it.

We educate computers by providing them with numerous instances of queries and solutions. The entire set of instances is referred to as the dataset by experts. A sample question(x) and its right answer are included in these instances (y). As an example, consider the phrase “a image of a cat” and the response “cat.”This is similar to the practice quizzes that some wonderful instructors provide before an exam. This practice question will help you prepare for the test by giving you an idea of what to anticipate. You may not get the exact same questions on the exam, but it will help you prepare for questions of a similar nature. We offer computers samples of questions and answers without instructing them on how to get the proper answers. The machine tries to find out a way to estimate the correct answer using the samples. Even if it hasn't encountered this exact question in the examples we've shown it, it learns to make an accurate estimate. The machine looks for similar questions it's seen before and makes a prediction based on the right answer it's seen before. That's all the computer is doing: guessing based on prior experience.

Computers do not execute Machine Learning on their own. In machine learning, humans play a critical role. Humans take on the role of coaches or facilitators, overseeing the entire game:

1. Ordered List Item Humans serve as models for computers to learn from.
2. Ordered List ItemHumans also instruct computers on how to recognize their errors.

A machine's sole purpose is to make educated assumptions based on examples and to correct those guesses to the best of its abilities. Everything else is governed by human-provided data and code.

All the machine is doing is guessing based on previous experiences with similar queries. (Prediction is a fancy way of meaning guessing.) It's not like you're guessing at random. It tries to learn to make the best prediction possible based on some interesting arithmetic (which we won't go into right now). To summarize, Machine Learning is all about figuring out how to make the best estimate possible. The closest approach to the correct answer is the best potential guess.

At first, the machine makes some extremely bad estimates (starting with random guesses). The program then compares its own guess to the right answer given in the example. We instruct the computer to make fewer predictions and make them more accurate. It then travels on a merry-go-round in an attempt to cut down on errors. It self-corrects as it strives to lessen its errors, bringing it closer to the correct guess. Experts refer to the process through which the computer corrects its own errors as gradient descent. There's no need to memorize the name any more. This might be the subject of a future entertaining post. For the time being, just keep in mind that the computer steadily corrects itself by looking at instances of correct replies.