Author Archives: Karhol Oleksandr

How was digital information stored previously?

Reading Time: 2 minutes

This might be a question that we only ask a little. It would be fair to claim that educated people know what a hard drive or a compact disk is. So there is no real need to investigate that subject more. Or is there? Well, it turned out that the era of storing digital data had its roots at the beginning of the 18th century. And the data holder was called a punch card. 

From Punch Cards to the Cloud — Google Arts & Culture

The history of punch cards began in the year 1725 when Basile Bouchon decided to automate the textile loom by making it follow a certain pattern without human assistance. The operator should still be present near the machine to operate it but not entirely. A punch card was a piece of paper with some digits. To give the input one should make a hole where the selected digit was. The card was then read by a reader column by column from above to bottom. Due to the very limited volume of data that could be held within one card, there were usually a pretty big amount of cards for one piece of information. Believe it or not but the principle remained the same even in the 20th century. The design of a punch card as well as the digits on it was adjusted according to the purpose. Some of them could store a binary code like 01001000 01100101 01101100 01101100 01101111 00100001 which says “hello”. Some others were readable for humans and used mostly in offices and archives to store demographic data. And the interesting fact is that punch cards were used by the USA at the end of the 19th century to conduct the 1890 US census. 

Another interesting case was the usage of punchcards for the SAGE air defence system that was expected to protect US citizens from a nuclear attack during the Cold War. The whole programme was stored in punch cards. The amount of data was only 5 MB while the number of cards reached 62,500. This created a very serious threat – the possibility of losing some or changing the right order. And of course, it was just not reliable to use paper cards as a key element of the national security system. 

Woman standing next to thousands of punch cards

The usage of punch cards started declining in the second part of the 20th century as new solutions came on the market. However, it is quite interesting that even I took the last high school exam by marking my answers in black so the machine could read them. Same method, but a new technology.

Reference:

  1. https://www.computerhope.com/jargon/p/punccard.htm#:~:text=As%20the%20card%20is%20inserted,written%20to%20a%20computers%20memory.
  2. https://en.wikipedia.org/wiki/1890_United_States_census
  3. https://www.redhat.com/en/topics/data-storage

Aircraft built like a cell phone

Reading Time: 2 minutes

We had a chance to experience the first time when the 6th generation military aircraft was presented to the public on December 2nd. The moment when Northrop Grumman’s B-21 Raider was unveiled at the company’s factory in Palmdale made many US foes hold their breath for a while. Besides an outstanding design and obvious aerodynamic characteristics, the aircraft has some hidden “gems” that Northrop Gruman shared in their report. This article will guide you through some exciting solutions that were implemented in the masterpiece called Northrop Grumman’s B-21 Raider. 

Inside the Making of the Military's B-21 Raider: Exclusive | Time

Cloud technology. Believe it or not, this is by far the first military project that managed to transfer all data to the shared cloud for common use by the US Air Force and Northrop Grumman. The details on what exact data is shared were decided to remain classified. However, having the fact that ‘This demonstration included the development, deployment and test of B-21 data, including the B-21 digital twin, that will support B-21 operations and sustainment.’ we can assume that a cloud-based solution was utilized for the development process and better data flow between the military supplier (Northrop Gruman) and client ( the US Air Force). The influence of cloud technologies may be observed after comparing the duration of development of various US military air vessels. It took 11 years to develop the D-21 Raider from scratch. While the development of the B-2 bomber (previous version) lasted for about 3 decades. This is quite a dramatic difference taking into consideration that B-21 is way more advanced than its “younger brother” B-2 which is rather to be called a grandfather. 

Open Architecture. B-21 Raider has updatable software which allows it to be flexible and agile when it comes to meeting the evolving threats of war or the rapid military development of the US enemies. This sounds crazy, but this bomber is able to update as your Tesla does. Engineers developed software that works as an operations system. Avionics, system controls, weapon controls, alarm systems, engines, and hydraulics are super fragile parts of every plane. It is known that the malfunction of one can lead to its further substitution and inability to fly. This issue was solved by Northrop Grumman engineers since software flexibility also means hardware flexibility. In case any problem is detected and the hydraulic system, for example, performs poorly on high attitudes, it may be solved even during a flight by changing the way pressure is applied to it. 

Hopefully, we could utilize such technology in the civil airline industry. Upgradability is a must nowadays. I truly believe we will be able to see such a message from our local airlines: “The flight to Dubai was cancelled since our Boeing 777 is undergoing a yearly upgrade”.

It was 1997 when the chess world changed forever.

Reading Time: 2 minutes

There is an old joke about chess – “When do we get an update?”. Well, despite the rules did not change in 1997, it was still a pivoting moment for all chess players. Gary Kasparov, a player who dominated the chess arena for about two decades, loses to a computer. As he said later on the Lex Fridman podcast: “It was physically painful to lose. It was not my first defeat against a computer, it was my first defeat, period!”. Deep Blue, a supercomputer developed by IBM, managed to outsmart the depth and creativity of humankind.  

Gary Kasparov vs. IBM's Deep Blue: Historic chess match before 'Queen's  Gambit' chess boom - The Washington Post

Nowadays, chess engines are nothing special. There are plenty of solid open-source solutions for all kinds of problems. You can train by playing a computer you can develop your own chess bots, and you can use free databases to monitor the recent trends in chess. 

Basically, chess engines are programs that calculate the best possible move at a certain position and predict the flow of a game up to tens of moves ahead. In order to give a bit of intuition here, we can compare the Elo ratings of top chess engines with top chess players. The Elo rating is a score used to mark the player’s level. So, the Elo of the current world’s best player Magnus Carlsen is 2864. While the rating of the most powerful chess engine is estimated to be 3000. It should be mentioned, that even a difference of 50 points is significant on that level. 

There are plenty of different engines that do the same job but have different features. For example, the Stockfish engine is utilized by chess.com for the analysis of your game. It gives you a list of your moves with an explanation of what could be done better and grades you play. It also gives you two important metrics – depth and precision. They are quite interconnected, and there is a positive correlation between them. Depth is the number of moves that the computer can calculate. Precision is a number that grades your overall performance. It can also show your best moves, moves that are recommended by chess books or your blunders. Komodo is another engine that is used by chess.com for bots. Whenever you play it, you play against a machine that is called Komodo. 

The use of machines has dramatically changed the way we learn, teach and play chess. It is another world where the games are perfect. This is also a reason why grandmasters do not always recommend using them – we can lose the creativity that used to be a decisive factor in a chess battle.

Sources:

  1. https://www.chess.com/terms/chess-engine
  2. https://www.youtube.com/watch?v=8RVa0THWUWw&ab_channel=LexFridman
  3. https://stockfishchess.org/

Mediapipe – made by Google

Reading Time: 2 minutes

Mediapipe makes Machine Learning faster, easier and cheaper. Basically, it is a versatile library developed by Google that is capable of understanding video and photo input. Mediapipe can detect faces, track your moves and gestures, find objects and know the dimensions of an object. Besides that, it may be used by developers of all kinds since it supports various programming languages. 

Quite an interesting area may be the way Mediapipe is built. It is actually a mix of other algorithms that are combined into one. Let us analyse one example. When Mediapipe detects your hand, it projects 24 so-called landmarks on it. Those are key points on your hand – joints. Those are points where you bend your fingers or simply rotate the hand. The Mediapipe algorithm is taught with thousands of hand images to detect the right landmarks on both moving and stable hands. Then, landmarks are put in a three-dimensional graph where they represent the location of the hand’s key points relative to each other. If a developer wants to use Medipipe for his own projects, then it is a great choice in terms of agility and efficiency. Having an output that is a list of 21 X,Y,Z coordinates is way computationally cheaper than having a list of every pixel of the particular image. 

Hands - mediapipe

If a developer wants to translate sign language to English written text, then it will need a huge dataset of pictures representing every sign. On average, only one data point would be a list of at least 2500 pixel values (for 50×50 images). Machine Learning algorithms would need to make expensive calculations with every data point having 2500 float values (minimum). However, it is possible to have 63 values within a data point with Mediapipe. No matter how big the image is. When dealing with a dataset of images, it can be easily “translated” into 63 coordinates representing the position of your hand. 63 coordinates is the number of landmarks (21) multiplied by the number of axes we use (X,Y,Z). At the end of the day, AI enthusiasts will have a small dataset that fully represents the hand gesture with a help of only 63 numbers which makes it really cheap to run a Machine Learning algorithm on such data. 

Hands - mediapipe

This outstanding approach allows us to keep the knowledge we possess, but make it more compact and comfy for further exploration or development. This is only one of 8 great features that Mediapipe has. And this is truly inspiring.

Link to the documentation: https://google.github.io/mediapipe/

Data analysis & visualisation for newbies

Reading Time: 2 minutes

While Python and R are commonly used for data analysis by IT specialists, it may be quite a heavy burden for those who just started their business path. Young entrepreneurs, as well as big corporations, are not always eager to run local data analysis departments due to the high costs of human resources. They use business analytics tools instead. 

There are multiple examples of such tools. Some of them operate as PaaS (Platform as a Service), and some as regular desktop apps. The most popular examples are Power BI, tableau and Google Data Studio. The main advantage is the relative easiness of use. A 10-hour course would be enough to learn how to create interactive dashboards with interactive business data. As a rule, such services have premade templates where a user may insert data. However, it is possible to create your own. Data importation is also not a big deal. Once you have a .csv type file, you import it as you would import a regular .csv in excel through data -> import .csv. 

And, of course, it is crucial to conduct the market positioning of business intelligence tools. If drawing a graph where the X axis is the extent to which the tool is difficult to use and the Y axis is the level to which you can get insightful outcomes, then BI tools would be somewhere in the middle. Excel would take the middle-down position, while Python and R would be in the right upper corner. 

Besides the easiness, BI apps are capable of visualising the data. All types of graphs that are commonly used by data analysts are available for BI specialists. Usually, it takes from 2 to 4 arguments to make a beautiful graph. The Y axis will take the column with numerical data – income, for example. And the X-axis will take age categories. It is also possible to insert a column with people’s gender to get two lines coloured differently to see the difference between males’ and females’ income within the specific age category. 

These all make business intelligence a perfect tool for agile and efficient data visualisation with the following decision-making process. 

Links for mentioned services: 

  1. Power BI: https://powerbi.microsoft.com/en-au/  
  2. Tableau: https://www.tableau.com/ 
  3. Google Data Studio: https://datastudio.google.com/ 

TikTok – How does it work?

Reading Time: 2 minutes
Image result for tiktok

The TikTok algorithms are subject to constant discoveries because of their uniqueness and innovativeness. They manage to capture peoples attention for at least 10 minutes after opening the app. While Instagram can not show off such results – only 3 minutes. This is a piece of strong evidence proving the great advantage given by recommendation algorithms rather than “like&follow” algorithms. So how do they work?

From the creator’s point of view:

After the video is uploaded the TikTok analyzes every kind of content related to this video (images, videos, text, hashtags). Algorithms capable of doing that helps creators to popularize content by recommending it to the global audience based on their preferences. This is thanks to the technologies provided by Bytedance Corporation, the company that deals specifically with algorithms for video-platforms such as TikTok or its Chinese counterpart Douyin. As a rule, TikTok analyzes the duration of the video, hashtags, theme, music used, location, time of publishing etc. 

From the user’s point of view:

Each of these factors is individually considered by TikTok’s For You recommendation system, which means that every For You page will be fully unique up to the interests of a particular user and his level of interest:

  1. Users interaction: liked/shared videos, followed accounts, the comments you post, and the content you create.
  2. Video you like/share information: This may include details such as captions, music and hashtags.
  3. Device and account settings: language settings, country settings, and device type.

For instance, TikTok reveals that a strong indicator of interest may be long videos (up to 60 sec) watched be a user. This will be more significant than a weak indicator, for example, the viewer and video creator are in the same country. Videos on the page “For you” are classified according to these “indicators of interest” and on the probability of user interest in certain content.

AI in trading – Trading robots

Reading Time: 2 minutes
Image result for trading

“There are only two emotions on Wall Street – fear and greed” – William Lefevre

Trading robots were created in order to eliminate the psychological element of trading. They analyze the market and open the position automatically without any help from human beings. Basically, it is a computer programme making its own decisions based on collected data such as the volume of the market, trend direction and trend power, graph patterns, fake pumps and dumps etc. 

Although, trading robots should not be regarded as a “holy grail”. It is because the market is way more unpredictable than most people may assume. Robots are seeking the rationalization of the price change, while the price is trying to be as tricky as possible. So most professionals use such robots as an auxiliary tool for comprehensive analysis of the market but not as a main decisive factor. 

It is worth mentioning that robots are quite adjustable. That means every trader may adapt the programme and tailor it to his own needs. For example, every trader determines the lower acceptable loss (the price where the loss is not so considerable) after crossing which the position should be automatically closed. This is due to the volatility of markets: prices are going up and down, sometimes it makes rapid movements but then returns to the starting position. To handle all these risks traders use the stop-loss tool. And of course, robots allow traders to settle exclusive rules regulating risk management. 

All the robots are written in MetaQuotes Language (MQL) which is a programming language used specifically to create automated trading robots and market indicators. That gives a great advantage for users because it means that every single person may not only buy robots but create them as well. 

Why do robots mistake? The answer is that usually they do not really predict, they rather interpret the chaos occurring in the market. So they do not outstrip the market they rather leg behind it. Robots try to find bonds between various factors and use algorithms to make a decision. But no algorithm (at least now) can predict that Elon Musk, for example, will decide to post a tweet #bitcoin and the price of BTC will rocket to the historical maximum just in seconds. Actually, no one is capable of doing that except for Elon himself)

Clubhouse – new approach to innovations?

Reading Time: 2 minutes
Image result for clubhouse

Let’s hit the trend and discuss the Clubhouse. 

The idea of Clubhouse is to make people gather in groups and exchange voice messages. Every group is thematic that allows users to find the appropriate circle of communication. This great advantage can be regarded as one of the possible reasons for the popularity of the Clubhouse. Another one is its artificial uniqueness. Social networks are full of memes regarding the invitation to the Clubhouse. There is a belief that it is a really complicated task to get invited. So basically the app is challenging human’s curiosity. 

I would say that it is quite a controversial approach to make people communicate. Dozens of users hate voice messages because they are not informative. Basically, developers of Clubhouse have taken a usual for everyone Messenger, deleted all functions except for voice messages and allowed users to choose a group according to their interest. And boom – the app is the most discussed one right now. The weird thing is that the idea is so old that it is difficult to believe in its huge success. Reddit offers almost similar service that is, as for me, more convenient. 

Another weird fact is that this innovation is basically a simplification of existing ones( Instagram, WhatsApp…). I mean that any other functions except for voice messages are not available. Is it really so weird? Actually not. Snapchat allows to exchange of photos and simple text messages, Tik Tok allows sharing videos and text messages as well. So simplification is a global trend in innovations. Apps are becoming more and more specific that attracts users. And that is the point of my blog. I want to discover whether we are going to keep using multitool apps, gadgets and all other tech products or using hundreds of simplified analogues. However, I have to mention that some other platforms like Instagram or YouTube are increasingly extending their functions. I mean YouTube has added stories and Instagram is a trading platform now as well. So we can see two trends: simplifications and expansion of functions. Which one will be overweight, that is the question. 

And of course, I do have to emphasize that Elon Musk and others have contributed to the success of the platform. Firstly the $12 million investment by Andreessen Horowitz Company then Elon Musk huge interest and some other celebrities have done Clubhouse a favour. So possibly that can be the key reason for its popularity. Basically, it is)

KERS

Reading Time: 2 minutes

Almost all means of transportation are developing in terms of speed nowadays. Trains are becoming faster and faster, car manufacturers are competing with each other in order to determine which car accelerates better, etc. Probably the fiercest competition can be seen during Formula 1 race. A normal speed exceeds 300 km/h. Although, the racing track is not straight. So racers have to use brakes and steering wheels so that they can do pivots. And sometimes the speed should be reduced from 320 km/h to 120 km/h in a second so as not to make a car crash into the wall during the turn. Afterward, the car needs a rapid acceleration. For this purpose, KERS is used. KERS is a Kinetic Energy Recovery System. 

Kinetic Energy Recovery System (KERS) is used for immediate recovery of the kinetic energy of a vehicle during rapid braking. The recovered energy is stored in a special reservoir for further use while accelerating, as a rule. The capacity of these reservoirs allows them to store up to 120 kilowatts for later purposes. 

The process is quite complicated and it is impossible to understand it without specific knowledge. But the diagram reveals the basic principle of the KERS. Definitely, the most important stage is the second one. The technology converts kinetic energy into an electric one. The reason why this should be done is that kinetic energy can not be stored and used as well, of course. The next step is to make a Direct Current (DC) from Alternating Current (AC). Basically, the purpose is that DC power has higher efficiency and size to power characteristics. The Storage stage implies collecting this energy. Afterward, it can be used by pushing a button on the wheel. Energy is discharged what adds more power to the engine.

This technology is not so widely spread. However, scientists try to implement it in various fields. It has performed successfully during F1 races and now KERS is being tested on trams, motorbikes, bicycles, and trains. 

 

Continue reading

Sheepview

Reading Time: 2 minutes

If Google refuses to make the Google Street view ….  sheep will. Citizens of the Faroe Islands,  one of the most beautiful islands in the world, asked Google to make a special service for users so they can admire local beauty online. First, their request has been rejected. However, locals have discovered another way to create online 3D maps.

The idea was to attract Google’s attention. So citizens attached cameras to sheeps’ bodies and let them go. In this way, animals have become real explorers helping to solve such a complicated problem. Animals have been having a walk through overwhelming landscapes, dreamlike paths, and green hills. In addition, images have been captured in various ways: by cars, kayaks, ships, on foot, and even wheelbarrows were involved. Almost every creature has been working on this project. The outcome has exceeded all expectations. Organizers believed Google to pay attention and finish their work. Actually, that has happened. Woolly pictures have become a local trend and now there is a specialized service that performs separately from Google. It provides users with unique “woolly” photos and videos, offers sightseen tours and accommodation facilities.

Moreover, locals have also made their own version of Google Translate – The Faroe Islands Translate. This is because you can not learn Faroese words and phrases using Google Translate. 

That is an example of how a problem could be solved in a creative way and how to monetize the outcome.

 

Sheepview 360 

https://visitfaroeislands.com/