Miki Szeles's Adventures in The Field of Data Science - The backstory
6 min read
In this article series, I will write about my adventures in the field of Data Science. This adventure just began a few weeks ago. Let me share my backstory so I can provide some context for you.
I was fascinated by AI since I was a child. I still remember how much I dreaded after seeing Terminator 2 in the cinema. I was fascinated and dreaded at the same time. The scene when the nuclear explosion blows through the city has been burnt into my mind forever. I had nightmares for years thanks to that.
I was always amazed by universes featuring artificial intelligence so Frank Herbert's and Brian Herbert's Dune (even if most of the story is in a post-AI world) and Isaac Asimov's universe completely enchanted me.
I have always dreamt of creating my own artificial life/intelligence. Later during grammar school, I met with Genetic Algorithms which is an amazing concept. You mimic the evolution to breed better and better solutions from generation to generation. Of course, this was far from being an artificial life, but I already had to use some "DNA" to code the properties of my artificial entities.
Years passed and I met with Artificial Intelligence and Neural Networks at the university. It was interesting but nothing really practical. But there was a movie called Paycheck featuring Ben Affleck and Uma Thurman. Predicting the future always fascinated me. It is still one of my favourite movies. I leave you a trailer in case you haven't seen it:
Whenever I started working at my first company, I learnt about Go. Which is a traditional Chinese game which conquered Eastern Asia and it is conquering Europe too these days. It is the Chess of the Eastern countries. Well, actually Chess is the Go of the western countries as Go is 4500 years old and compared to that Chess is a youngster only 1500 years old. The difference between the essence of the games reflects well the difference between eastern and western thinking. While in Chess you have to defeat your opponent by killing the king, in Go, you split up territory so you can live together.
I immediately fell in love with Go. I found a Japanese anime called Hikaru No Go which really boosted my motivation. Check the trailer for yourself:
Wow, I was looking for a trailer and I just realized there is a Hikaru No Go remake.
In the beginning, I was afraid of playing online so I started with an AI-based game and played a lot with that. I improve a lot, but there was no real variety in the play of the AI. So I moved to online and offline gaming after a while.
In the beginning, it was quite disappointing. I fell into traps of simple tricks like the ladder. So in order to improve, I started to read books and also practised Tsumegos. They are go problem based on life-and-death which means you have to make the right steps to capture your opponent's stones by encircling them or make the right moves to save your stones. This was a really fun way of learning.
I reached around 10 kyu, which was nowhere compared to the elite. the amateur league is starting at 30 kyu it goes to 1 kyu then there are 7 amateur dans. Being ranked at an amateur level is already a spectacular achievement. In that time Diána Köszegi and Csaba Mérő were the best Hungarian players. Diána was at 7 amateur dan and Csaba was at 6 dan. I tried to google where they are at the moment, the latest info that I found is Diána Köszegi being a 1 dan pro and Csaba being a 6 dan amateur.
Ok, but why did I talk so much about Go? During the years I heard there are artificial intelligence based players which were already able to defeat the best Chess player in the world. Deep Blue running on IBM's supercomputer defeated Gary Kasparov in 1997 just one year after Kasparov defeated the AI in 1996.
As Chess was solved, AI developers looked for new challenges. A straightforward decision was to turn towards go, as it has a much higher complexity as in each step you have to put a stone on a 19x19 board which is much more options than in Chess. For a newbie the 19x19 boards are very scary cause it really means more than 300 potential moves in each turn, but as you advance you can narrow down the potential moves to a few.
Go players are fascinating, they can recall complete matches from their minds. As far as I know, most of them do not have a photographic memory, but they remember the sequence of moves of their moves and their opponent's moves.
It took another 20 years in AI advancement to beat Ke Jie who was the best Go player in the world at that time. Google's DeepMind AlphaGo was the winner. You can read more about AlphaGo on Wikipedia.
So as Go was solved, game AI developers turned towards more complex real-time challenges with endless potential moves and imperfect information.
It took only 2 additional years to develop and train DeepMind's AlphaStar AI to defeat 99,8% of the best human Starcraft 2 players. In case you are interested how AlphaStar is trained read this article.
After this, I did not follow the progress of AI game players. I bought some Udemy courses about AI development but I never had the time to delve into the topic. Up till I saw a post in a Facebook group.
A Data Analysis class started in a few weeks, which were not only available for students but they also accepted external students. I immediately got excited about learning about AI, BigData and predicting the future.
The course took place at the Budapest University of Economics and Technology which is my alma mater institute. I got excited to be again between its walls.
So I registered for the course and I had to wait whether I got accepted or not. Days passed and no feedback come but in the end, 3 days before the class started, I got the email that I was accepted.
There were lots of administration, registering to different websites. I was quite happy as I got back my old Neptun code. There was only one problem. There was no online option to change my password, so my only chance was to remember what password I used more than 20 years ago. I managed to remember. 😊
The first class started on the 15th of February which I could not attend, as I got COVID but thanks to my (soon to be) Isabella I have the recordings from that class.
The class is educated by Data Analysis professional from DMLab. They have more than 15 years of experience in the field with many successful projects. It is fascinating to learn from such professionals.
So that was my backstory, I hope you did not regret reading it.
Let the adventure begin... 😊
In case you do not want to miss my posts, just follow me here on Hashnode, on LinkedIn, on Twitter on Medium.com, on dev.to and even on Instagram. 😊
Did you find this article valuable?
Support Miki Szeles by becoming a sponsor. Any amount is appreciated!