Blog and articles

This section contains writings about machine learning, quantum computing and cryptography.

Compromised privacy

In this article we consider privacy and security (P&S), IT systems vulnerabilities and possible outcomes. Blog about Where is my privacy?

Machine learning as a competitive advantage

I got a question what is the next disruption for the software development and I put together some of my thoughts. Blog about Machine Learning – The next big game changer.

Article about quantum cryptography

In this article we discuss about phase manipulation in quantum cryptography and phase gates in quantum computing.

Simplify complex problems with machine learning and analytics

A guide to simplification - think before act.

Turbulent water jet (Van Dyke 1982). Photograph P. Dimotakis, R. Lye and D. Papantoniou.

Turbulence is one of the oldest and hardest open problems in physics. Is a turbulent flow fully chaotic, is it integrable or is there some fine-structure in the flow? Does there exist a machine power which can calculate a solution or a partial one? How to even approach this problem? A simple question whose answer becomes quickly hard. (In a simplified case an analytical solution based on field theory can be found showing the scale-invariance.)

This example has two characteristics we are talking about here:

  1. How to simplify a complex problem and to find the main dominant behavior from a large data set.
  2. When considering a hard problem for a while, then other tough-looking problems turn out be easy and become solvable. Everything is relative.

Search for small data

Today we get lots of data from different sources and machines generate that even more. The challenge is we have even too much data to be analyzed, and to find the essential or trends may require a lot of human work and machine capacity.

Our aim is to use intelligent models and algorithms to minimaze the amount of work and data to find significant results. Instead of using "big data" we go with the search of "small data". This is made through identifying the weak signals, using the most valuable data sources and justifying the prominent movements which determine the main behavior. In many cases most of the data is irrelevant in the context of the topic that is investigated. Additionally, what is great at the time we are living, there is free data available everywhere and also great sophisticated programming tools and libraries which make the data processing relative easy. The objective is to utilize these capabilities as much as possible, and to use machine learning algorithms to automate the processing and to make accurate future predictions.

In this blog or site we will go through these principles with examples. Please enjoy.

Back to the main page.