fbpx

Machine Learning 101 with Kiril Cvetkov

So far 2020 has been a striking year when it comes to Data Science.

Go through the Machine Learning 101 with Kiril Cvetkov who designed the Data Science Bootcamp prep programme, solving all beginner problems to those who don’t know the basics.

The fall intake for the Data Science Bootcamp at Brainster is almost full. The new participants are already taking part in the prep programme, designed to solve all “beginner” troubles of those who don’t know the basics. However, we didn’t stop here. We are continuously strengthening the team of instructors and curriculum contributors to deliver a once in a lifetime learning experience to the Bootcamp participants. Another top-notch practitioner with more than 6 years of experience in the field of Machine Learning and Data Science is joining the team of curriculum contributors in the Data Science Bootcamp.

If you are looking to kick-start a career in Data Science, make sure you go through our Data Science Bootcamp Curriculum. The programme is designed by experts with more than 15 years of experience in the industry as well as academia. 

Kiril is a Machine Learning Engineer at Symphony.is with a great passion for Computer Vision and Deep Learning. He works on projects that have more in common with the sci-fi sphere than the reality we are used to. These technologies are leading the evolution of Data Science and Digital Transformation in general. He is one of the curriculum contributors to the Data Science Bootcamp organised in Vienna. You will be able to listen and exchange opinions with him on several occasions throughout the Bootcamp. Naturally, Kiril will always be here to help you out in the process of learning.

We had a chat with him, that could be interesting for anyone interested in Data Science and Machine Learning. 

Brainster: Kiril, Let’s go straight into the “deep” – Keras, Tensorflow, or PyTorch? Go!

Kiril: Keras is usually subject to Tensorflow and gives us a high-level API abstraction of various neural layer implementations with the primary purpose of building a fast prototype. Combined with Tensorflow, it also supports the ability to create a variety of neural architectures and models to delve deeper. I personally use it when I want to try something in a fast.

Tensorflow, on the other hand, provides greater flexibility to penetrate “microscopically” all segments of neural networks. It is exceptionally fast, efficient and optimized for working with big data. However, the principle of “dataflow” programming made my debugging and debugging process difficult. Therefore, the code can often become chaotic and unreadable.

Personally, I find myself mostly working in Pytorch. I like the principle of designing and developing neural network architectures, where even the code design fits in with that of traditional software engineering. It is easier to debug, errors are found, and tests are easier to write. Unlike Tensorflow and Keras, in Pytorch we type more code, but the benefit of that is that it can easily be kept tidy. I recommend it for bigger and long-lasting projects.

Brainster: The trend and application of Deep Learning continue to grow daily. According to many experts in the field, this is where the future of Data Science lies. How much do you share this opinion?

Kiril: Every day in the field of neural networks we have new inventions and new improvements in the solutions of various problems. Each of those inventions gets a different form of “Lego cube”, which can fit and improve another problem. Deep Learning is a branch that is becoming more and more independent, hence I believe and see a great future in it.

The admissions for the next batch of students on the Data Science Bootcamp are open. Save your spot now and begin our online prep programme.
Machine Learning 101
Brainster: You are one of the authors of the Data Science Prep Programme, and curriculum contributor of the Bootcamp. A quick look at the material – from unsupervised and supervised learning, through Python, NumPy, Matplotlib, regressions, and neural networks… This is a realistic fast-track to the key Data Science basics. Why did you decide to structure the course like this?

Kiril: My idea was to start with an introduction and lectures on topics I “fell in love with” when I started working in the field of Data Science. I want to share my passion with other people and thus motivate them to continue working in the field. We will learn the basics that will help everyone in the future to solve practical problems by getting to know the necessary components. This will help you understand how the “brain” works in the virtual world.

In the course, you will learn the basics in this area – the two key differences in learning concepts in the world of artificial intelligence. You will get acquainted with the introductory commands in Python – the most used programming language for solving problems of this type. Numpy will help you learn about dimensional data manipulation while for visualization and presentation you will work in Matplotlib.

Through regression and neural networks, students will be able to learn how to use data to predict and recognize any relations as well as solve practical problems.

Brainster: Tell us something more about yourself. Having in mind your beginnings as a software engineer, what was your career path in Data Science?

Kiril: I am interested in software issues where you can apply natural thinking, algorithmic skills, a certain level of mathematical theory and the ability to solve a problem that has not been solved well enough so far. Thus, I came to the path of data processing and analysis, neural networks and agent reinforcement learning. In my case, the desire to learn and passion was enough to enter the field of Data Science. Not to mention the perseverance, dedication and time spent around the matter.

Brainster:  Through an off-record conversation, you mentioned that much of your Data Science knowledge you have acquired yourself. How did you do it? Having your experience in mind, what do you think is needed for quality education in the area of Data Science?

Kiril: Today we live in a world where the Internet gives us the opportunity of various connections, knowledge exchange with other people in the open-source world, reading scientific papers and posts, etc… Therefore, I concluded that you can learn what you want and at the same time cover the areas that interest you personally. Of course, it takes time.

Everything has its beauty and in my opinion, if someone wants to monetize their knowledge in business or use it to create a product, you should dedicate time to self-promotion, work and education in a particular community. On the other hand, if you want to become a scientist and work on research, you should take on the academic side. Those who are more ambitious can balance both sides.

Brainster: What does your workday look like? What are you currently working on?

Kiril: I am currently committed to several projects. I work as a Machine Learning Engineer at Symphony.is where I am surrounded by wonderful people and a pleasant community daily. I am also writing several scientific papers to be published soon that are in line with my postgraduate studies. Also, I am working on my own “start-up” application with another friend, which fully uses AI and should be released very soon for iOS. Sounds too much, but if time is planned properly, it can be done.

Brainster: In most digital skills, things change around the clock. In digital marketing, for example, a small change in Facebook’s algorithm is enough to make your whole strategy fall apart. What is the case in ML and Data Science? Tensor Flow 2.0 was received with applause, but do “updates” sometimes cause you problems?

Kiril: Updates are the same everywhere, not just in the IT world. The new version of a computer or phone operating system, new trends in music, fashion or show business. Some people will like it, some will not, some can be influenced positively, others negatively. As time passes by, most often each change corrects the lack of the predecessor. I think that every change should be welcomed and deserves to be tried and tested. If it is really worth it, we should be flexible and accept it. Naturally, getting the most out of it.

Brainster: Kaggle and LinkedIn have long argued that while there are plenty of open positions for entry-level Data Scientists, there is a real shortage of senior staff with years of experience. Is 2020 a critical year when we should start training for Data Science?

Kiril: A decade ago, businesses worldwide could not identify this field as it looks today. They were just no work around this area. However, times have changed, businesses have realized how much Data Science is needed to target groups, predict trends, minimize costs, or use artificial intelligence to create applications that will enhance or make a person’s everyday life. more interesting.

So now, we are surrounded by open positions of this kind across the globe. Today I can proudly say that Data Science is the “hottest” job position on the planet.

Machine Learning 101
Brainster: For these reasons, AutoML (Automated Machine Learning) is an explosive and controversial topic right now. What do you think about this new trend in Data Science?

Kiril: This awaits us sooner or later. Everything that is done today and has given a good result will be automated and tomorrow can be used even with the click of a button on the interface.

But why not? Imagine for a second you were transposed into the karmic driven world of Earl. Through such a tool, even visualized, he will be able to make revolutionary discoveries in his scientific field.

The same goes for other natural areas such as biology, physics, astronomy, chemistry, and even some of the social: business and economics. Not to forget the automated processes that took place in software engineering. They revolutionized and helped developers and the IT community produce software much faster.

ML automation will be able to help people whose knowledge is subject to various natural and social branches in the world. I is even possible to stand out as a separate scientific branch. Each automated process opens up opportunities for new work and research.

Brainster: The Data Science Bootcamp had a record number of applications in 2020. These participants will spend a significant time on the prep programme. What is your message to the current and future students at the Data Science Bootcamp?

Kiril: It is worth trying what is new and what is unknown. We need to have a clearly defined goal that we want to achieve. If you find yourself in what you will learn, it is really worth investing time and effort in it, where the price is getting out of your comfort zone and your routine habits.

When you taste success, you will be happier and fulfilled because you have found yourself and succeeded in your idea.

Machine Learning 101

If you are looking to start a career in Data Science, the Brainster Bootcamp will get you covered. Check it out, it might change your life.

Share on facebook
Facebook
Share on twitter
Twitter
Share on linkedin
LinkedIn

Leave a Reply

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