Basics of Machine Learning (for Beginners and even Older people)
If you are reading this blog probably you have some basic knowledge about science, technology, engineering, or mathematics, but does not matter if you do not have, below I am going to try to explain some concepts about Machine Learning in an easy way (if you have doubts after read whole blog, do not hesitate to contact me or write a comment and I would try to clarify them for you).
I think that the best form to understand something which seems very complex (which maybe it is not) is first of all try to split in the big concept in more little parts, so first we are going to understand the concept of machine, — according to oxford’s dictionary, is an apparatus using mechanical power and having several parts, each with a definite function and together performing a particular task. — so for example a typewriter (if you are an old-fashioned writer), a juice extractor, a fridge or even a fancy retractable pen could be called machines.
Now let’s talk about the Learning concept, next there are some popular definitions about it:
- The acquisition of knowledge or skills through study, experience, or being taught.
- The process of acquiring new, or modifying existing, knowledge, behaviors, skills, values, or preferences.
- Knowledge or skill acquired by instruction or study.
- The act or experience of one that learns.
Then having those concepts, we can say that Machine Learning is a scientific discipline that creates systems that learn automatically, for example, a typewriter that is able to understand what you type and suggest words or paragraphs to complete your writing obviously with a perfect grammar, or perhaps a juice extractor that is able to receive your health status directly of information supplied by your doctor and based on that prepare a juice with fruits or vegetables that could help you to stay healthy, but always taking into account your favorite tastes.
At this time, you may have noticed that Learning in this context requires identifying complex patterns in millions of data, which means data is an important point to Machine Learning, after all, it is the basis for learning, what really makes a machine to learns are the algorithms that reviews the data and are able to predict future behaviors. Automatically, also in this context, it implies that these systems are improved autonomously over time, without human intervention. In other words, and comparing it to agriculture, the seeds are the algorithms, nutrients is the data, you would be the gardener and plants would be the programs.
Some people believe that Machine Learning will do lazy people, but also I believe that it could be very helpful for most of people, currently we have been seen some advances around the Machine Learning and some of them maybe have had expected applications, like Speech Recognition (Natural Language Processing in more technical terms), computer vision (Facial Recognition, Pattern Recognition, Character Recognition), or Self Driving Cars, but there are others applications unexpected like Product Recommendations (for example in Amazon, YouTube or Netflix), Data Mining / Big Data (basically are just manifestations of studying and learning from data at a larger scale), Stock Market/Housing Finance/Real Estate (All incorporate a lot of Machine Learning systems in order to better assess the market, namely “Regression Techniques”, for things as mediocre as predicting the price of a House, to predicting and analyzing stock market trends).
Now that you have a bit more information about Machine Learning let’s to go few deeper about how it works.
How does it work?
The main objective of every learner is to develop the ability to generalize and associate. When we translate this to a machine or computer, it means that they should be able to perform with precision and accuracy, both in family tasks, as in new or unforeseen activities.
And how is this possible? Making them replicate the cognitive faculties of the human being, forming models that “generalize” the information presented to them to make their predictions. And the key ingredient in this whole issue is the data.
In fact, the origin and format of the data is not as relevant, since machine learning is capable of assimilating a wide range of data, which is known as big data, but it is not perceived as data, but as a Huge list of practical examples.
We could say that their algorithms are mainly divided into three broad categories: supervised learning, unsupervised learning and reinforcement learning.
Kinds of machine learning
Supervised learning
It depends on previously labeled data, such as the one that a computer managed to distinguish images of cars, from those of airplanes. For this, it is normal for these labels or labels to be placed by humans to ensure the effectiveness and quality of the data.
In other words, they are problems that we have already solved, but that will continue to arise in the future. The idea is that computers learn from a multitude of examples, and from there they can do the rest of the necessary calculations so that we don’t have to re-enter any information. Examples: voice recognition, spam detection, handwriting recognition, among others.
Unsupervised learning
In this category, what happens is that the algorithm is deprived of any label, so it does not have any previous indication. Instead, it is provided with a huge amount of data with the characteristics of an object (aspects or parts that make up an airplane or a car, for example), so it can determine what it is, from the information collected. Examples: detect morphology in sentences, classify information, etc.
Reinforcement learning
In this particular case, the basis of learning is reinforcement. The machine is able to learn based on tests and errors in a number of different situations.
Although it knows the results from the beginning, it does not know what the best decisions are to get them. What happens is that the algorithm progressively associates the patterns of success, to repeat them again and again until they are perfected and become infallible. Examples: automatic vehicle navigation, decision making, etc.
There are other more complex approaches to more specific tasks, but it is not worth delving into them. At the moment we don’t want to complicate things.
Scope of Machine Learning
Many activities are already taking advantage of Machine Learning. Sectors such as online shopping — haven’t you ever wondered how the recommended products are instantly decided for each customer at the end of a purchase process? -, online advertising — where to place an ad so that it has more visibility depending on the user who visits the web — or anti-spam filters have been taking advantage of these technologies for some time.
The practical field of application depends on the imagination and the data that are available. These are some more examples:
- Detect transaction fraud.
- Predict failures in technological equipment.
- Anticipate which employees will be more profitable next year (the Human Resources sector is seriously betting on Machine Learning).
- Select potential customers based on social media behaviors or interactions on the web.
- Predict urban traffic.
- Know what is the best time to post tweets, Facebook updates or send the newsletter.
- Make medical prediagnostics based on patient symptoms.
- Change the behavior of a mobile app to adapt to the customs and needs of each user.
- Detect intrusions in a data communications network.
- Decide what is the best time to call a customer.
The technology and the data are there, the Machine Learning could be used to new ways of making decisions based on data, because as we saw, nowadays the data is very important and as some people said is the petroleum of the future.
Expectations about the future of Machine Learning
Well, since the conversations and comments of an endless number of digital consumers — whose number day by day continues to increase — offer these types of technologies an overwhelming amount of information, they continuously gain new knowledge and detect trends faster than any Human could do it.
While it is true that this enormous amount of data will make it much more efficient, it will necessarily require a lot of human talent to improve, since finally computers do not have such a high command of the language applied to reasoning. Or what it is, they are not exactly skilled at determining contexts.
Which means that for machine learning to develop in these areas, experts in each field of work will have to take the time to train the machines and gradually incorporate them into each of the processes they wish to refine.
Finally, as with all technologies, businesses will have to start by understanding the basic principles of this technology, in order to use it in their favor and improve the productivity of all their business operations. For now, it is estimated that this — like many other derivatives of AI — will completely transform the world as we know it.
As you will see, this was just an introduction to the intricate world of machine learning. In an era where innovative technologies emerge every time we blink, it is easy to get lost in the flood of information and new concepts.
References
https://machinelearningmastery.com/basic-concepts-in-machine-learning/
https://www.zendesk.com/blog/machine-learning-and-deep-learning/
https://www.youtube.com/watch?v=f_uwKZIAeM0
https://en.wikipedia.org/wiki/Machine_learning
