On July 23 Garmin services stopped unexpectedly.

Many think of Garmin as a fitness watch manufacturer, but the company has many more services to offer — most notably in marine and aviation range. As Garmin went black the services we shut down as well; many pilots and captains were unable to download and update maps, with little to no information from Garmin itself.

I’m not going to speculate about the reasons of blackout (which, as some sources say was caused by Wasted Locker ransomware attack), and huge recovery time (four days to put services online in a limited mode, and still limited as for the time of writing — total six days after blackout). …

And that’s one of the most elegant productivity techniques I’ve seen.

Computers are multi-functional. That’s their benefit. Is is also a biggest threat to the productivity for people who mostly work on a computer, as you can switch from being super productive and focused to browsing cat videos for an hour in a matter of keystrokes. Switching back, however, is hard, as we all know.

Factory workers, delivery guys, and other not-so-digital employees don’t have this problem — it’s kinda hard to get into Netflix marathon when you are on a 10 hour shift in a uniform and your hands are kinda busy. Many productivity techniques try to mimic these limitations. Pomodoro promotes kitchen timers, which are loud and red. Some authors recommend to change clothes, making your work and leisure times different. …

Basic income concept shows moderate to excellent results, but why it will fail on scale?

Modern society, while providing and individual with endless possibilities, exposes a lot of uncertainties — especially when it comes to living, transportation, education and medical care. In modern capitalistic countries and average person is left with a bit of monthly salary after covering all the leasing, insurance and mortgage payments. It’s OK when you have a nice job with respectable salary, but a lot of people doing service jobs have income right above survival level. …

Five simple and verified ways to make eye-poking, face-melting designs, explained.

Believing that you should like the design

Designing stuff is a quite technical process, that requires little creativity and enormous attention. Appealing to your personal likes / dislikes, or even to third persons like colleagues, relatives and co-workers can easily disrupt streamlined process and fall into bottomless hole of comparing numerosu opinions.

How to avoid that? Prefer value (the ways the end user actually can use an element) over personal preferences.

Assuming a design can be finished

A good design emerges from user feedback. Assuming that you can spend month over your UI, presentation, or application will most probably lead to bad navigation and over-engineered features. …

DIY AI continued: our first ML framework, FANN, explained.

Honestly, it is a bit challenging to write again after I’ve published Mostly complete chart of Neural Networks, explained — 150+ recommends and 200+ followers bring up some responsibility. Still, let’s continue the journey into Machine Learning, and transit from hardcode coding I’ve mentioned in Old-school matrix NN to Machine Learning frameworks. Pull the code from github repo, consult installation manual for Python / FANN and let’s code some NNs.

Speaking short, 100+ lines of code in numpy is reduced to nine lines using Fann:

Nine lines, clean code and much faster results comparing with manual implementation. Let’s dig a bit deeper and walk through all steps that are needed to feed the semeion dataset to FANN. …

The zoo of neural network types grows exponentially. One needs a map to navigate between many emerging architectures and approaches.

Fortunately, Fjodor van Veen from Asimov institute compiled a wonderful cheatsheet on NN topologies. If you are not new to Machine Learning, you should have seen it before:

In this story, I will go through every mentioned topology and try to explain how it works and where it is used. Ready? Let’s go!

A short time ago, Facebook developed bots that were able to bargain, they started talking weird, Facebook developers probably giggled and re-wrote the code.

Then, the Internet exploded. TechTimes write about “stopping bots before they evolved to Skypet”. NYPost called it “creepy”. CNet mysteriously wrote that “Facebook declined to comment”. BGR wrote a hart-breaking story about panicking engineers.

To be short, internet started buzzing (and it is a shame that big news agencies joined the buzz) about something like that:

When what really happened was that:

How do NN work and how to create and train NN like many years ago. Part of DIY AI series.

Pre-requisites & code

You will need Python and Python IDE. Consult DIY AI installation manuals if you need to install them (you will need Python and Console parts for this story).

You can download the full source code from Github here.

The draw.io diagram is accessible here.

So, how do neural networks really work?

As I have mentioned in previous article, since early 50s we tried to mimic the human brain and a smallest part of it — Neuron — itself.

The closest thing that acts like Neuron is a Perceptron — which is composed of multiple inputs, summing engine, single output and an activation function. …