Filip Jašek is a great enthusiast not only in Artificial Intelligence, but also in technological innovations and development. Besides neural networks, he enjoys climbing, nature and cooking. He started in nvias as a lecturer and mentor in several workshops (e.g. Minecraft). Gradually he started creating materials along with preparing programs and tutorials for more advanced workshops on AI (Artificial Intelligence). At the same time, he is studying for his bachelor’s degree at University of West Bohemia of The Faculty of Applied Sciences in Cybernetics and working on his other projects.
What is nvias and what projects are you working on
Nvias is a non-profit organization that handles projects announced by the European Union and with which we, Archetix , are partners. Their mission is to turn consumers into creators, specifically through workshops and other education in technology, business and practice. At the same time, the workshops are tailor-made (mostly for 2nd grade students in elementary and high school), so it’s much easier for students to adapt and learn new things. One of the main goals of the organization is to introduce and motivate the young generation to use and work with new technologies. In this way, the students of nvias will gain a much deeper insight and knowledge about the new developments in the technical world. Ideally, nvias will help them find the impetus to create or further their interest in this field.
Nvias is also active among universities, specifically with the University of South Bohemia in České Budějovice, where the topic of artificial intelligence was of interest to PhD students. In this case, the lecturers at nvias again adapted their educational materials for the audience in question.
On the same basis, nvias also educates companies in the field of AI through an entertaining format.
Nvias cooperates with the University of West Bohemia in Pilsen with the FAS, which helps them in the framework of implemented workshops such as the ongoing AI Discovery Show (see more about the workshop below).
What is the AI Discovery Show
The AI Discovery Show is a follow-up project to the already established AI Discovery workshops. AI Discovery is a fun workshop format that introduces students to the core issues of AI in four hours. Unlike the one-day workshops, the AI Discovery Show is more extensive. There are a total of six 2-hour workshops and a final show. There participants have the opportunity to present their projects. Students are divided into several teams and work on their projects. They always receive feedback for their project from individual nvias tutors or practitioners (e.g. Ing. Jan Švec , Ph.D.). The event will take place on 12.12. 2022 at 3 pm at the SIT Port and participants have the choice of online or offline presentation formats. The final presentations are of course open to families, friends but also to people interested in the topic of AI.
This year, registration is already closed, but if you are interested in the workshop, either for you or your loved ones, you can certainly look forward to future events. Nvias runs very similar events such as the AI Incubator.
What is Artificial Intelligence (AI)?
Nowadays, Artificial Intelligence (AI = Artificial Intelligence) is taken as a marketing tool. It has become a concept that no one fully understands, but everyone admires to some extent. No matter where you look today, there is always a product that is enriched with AI that does something, manages something, answers something, helps e.g. to write, or the product is so called AI powered.
But artificial intelligence is a very broad term. We can see it already in sci-fi movies, when AI takes over the whole world or a humanoid/android (e.g. Terminator) saves or destroys the world. The whole topic of AI is somewhat “bloated” and involves things that have already been invented and have been working for maybe more than a decade (e.g. regression curve interleaving). So it is important to remember that AI is a very broad term that includes algorithms that have been around for many years and are generally nothing groundbreaking.
AI can be decomposed into sets consisting of Machine Learning and Deep Learning (see figure below).
Subsets of Artificial Intelligence and Neural Networks
Under the subset of AI is Deep Learning, the latter deals with neural networks, which have developed quite rapidly in the last decade. Deep Learning was already introduced and has been known for a very long time (1943). However, in the last few years it has been possible to develop rapidly in this field. Since we already have hardware that has the necessary computing power. And it is possible to get somewhere without waiting several generations to compute something. A lot of momentum is now with the advancement of technology, where even on a simple laptop or mobile phone (through a server such as Google) you can try and train a neural network (see image below).
What do we mean by a neural network?
A neural network takes its inspiration from the similarity of the neuron, a fundamental element in the human brain. There has been a long-standing phenomenon to try to mimic humans, both in similarity and in the way they think. A neural network consists of so-called layers, and each layer contains a number of neurons. One neuron takes inputs, then specific simple mathematics takes place in that neuron. That is defined by a function, and from that a certain output is derived. Which is distributed to other neurons in the next layer. Neural networks are made up of thousands of neurons. In fact, the output of a neural network can be anything. For example text generators, story generation (GPT-3) or image generation are very trendy nowadays.
There are several models and forms of neural networks, depending on how they are approached and how they are structured. One can use so-called frameworks, which are given structures such as TensorFlow. Where different packages are available to prepare datasets or to create the neural network itself. This means that we don’t have to write functions individually anymore. Instead they are already pre-prepared “templates” where we just insert layers. Thanks to the framework, many more people have access to try and create their own neural networks.
A cool web tool from Google is Teachable Machine. It is a free application that teaches your computer to recognize your own images, sounds, and positions. It’s a quick and easy way to try out and get familiar with Machine Learning models.
How is Deep Learning different from Artificial Intelligence?
It’s important to highlight that Deep Learning is a subset of Artificial Intelligence, and thus AI is Deep Learning, but it’s far from being all about it (see image of AI set). Artificial Intelligence is defined by an algorithm that can learn something, which is a really broad definition. One of the definitions that describes AI more closely is:
“Artificial intelligence is the simulation of human intelligence processes by machines, especially computer systems”.
Ed Burns, 2022
Which can be taken to mean that artificial intelligence is nothing more than the simulation of human intelligence processes by machines, especially computer systems.
