Neural Networks / Deep Learning Algorithms


“Graphics are 100% generated by an algorithm in one shot. No edit or post-processing. I used videos recorded from trains windows, with landscapes that moves from right to left and trained a Machine Learning (ML) algorithm with it.”

Mesmerizing project by Artist Damien Henry with deep learning / neural network algorithms. And there is also a very interesting paper about neutral networks that reproduce what our brain sees,  I just stumbled upon via NerdcoreSharing deep generative representation for perceived image reconstruction from human brain activity.


Generative Dream

After a few self-experiments, I came to the conclusion that no visualization comes close to what we experience in dreams. Also when sharing dreams, the perception of dreams is individual for all of us. However it is interesting that even blind people have sort of a similar perception of the world as people with sight. That’s probably the case because of story telling and describing the world to blind people from people with sight. For example: A sun has to be round and is shining.

A interesting investigation from graphic designer and artist Tobias Gutmann shows that a lot of people are stuck with a conventional representation of the world. He asked people from different cultural background to draw a house and a world. The similarities and differences between people from different background and especially kids are impressive!

You could say that the perception of the world is always very individual. Especially when compared with kids, people from non-western countries or even computers.

The same counts for dreams as dreams are fractals from something we maybe have experiences or not and put together in a way that most of the time doesn’t really make sense. You could go further and say that as soon as you share a dream, it is no more your dream but now the dream of someone else. Because this person is looking from a different angle on your dream interpretation.

I set myself the goal to find out how we can generate a (kind of) language for dreams with the help of generative art forms through programming.

After a lot of experiments with different visualizations code examples from the web and math formulas, I found a tutorial from tsulej describing folds and fractal flames and how to handle them in processing. Through variations you can generate flavour in fractal flames that generate beautiful, mysterious objects and drawings. I found a lot of similarities to dreams in those drawings and decides to investigate more time finding out how they work and how I could implement them into my dream project.




Analyse Brain to trigger Events

I did further research in how something like the Dreaming Buddha could be triggered more accurately. I looked how in scientific fields sleep and dream is analysed. With the help of EEG and EOG it would be way easier to trigger events as playing sound or soft vibrations on your arm in the right sleep phase.

I compiled some experiments but didn’t try them out myself as in my opinion this is a very scientific approach to dreams and rather tries to explain how our brain works. Also there are (or have been) already a few products that go exactly into this direction.

Feel free to conduct your own experiments with those two methods.
1. EOG/EEG Wearable or through open-API Wearables (Fitbit, Jawbone). Find both method sheets here:  EOG-EEG-API-to-trigger-events.


Alt Sleep Patterns

While dreaming is a phenomena that can actually happen at anytime, we still connect it particularly to the nighttime when we are asleep. I started a quick research into what other sleep patterns exist and wanted to find out how we can actually dream even more often.

Most of us sleep in the Monophasic sleep cycle. It consists of sleeping once per day, usually for between 7 and 9 hours per night. The monophasic sleep cycle is still a product of the long work hours of the industrial revolution that has remained a cultural norm even as work hours have shortened.

Before the industrial revolution most of people went to bed when the sun went down. This pattern is called Segmented sleep and is the most natural sleeping pattern according to scientific literature. It consists of two sleeps, both at night time, first going to sleep at dusk, and secondly waking at dawn, synchronized with the local lighting patterns, sunrise and sunset. Sleep is between 6 and 8 hours a day.

Then there are also very extreme sleeping cycles, like the Uberman, that consists of only naps to the point of getting rid of the core sleep entirely. While this sleep pattern is extremely difficult, it can have great benefits by increasing the amount of time in a person’s day. An Uberman will have 6 or 8 x 20 minute naps a day, with total sleep time ranging from 2-3 hours a day. Even more hardcore is the SPAMAYL (Sleep Polyphasically As Much As You Like) cycle.

SPAMAYL was introduced by Rasmus, and it seems that he is the only person to sit through this pattern in long term. SPAMAYLer can expect to take no less than 7 naps a day, and often need as many as 10. Rasmus usually get between 2.5 and 4 hours of sleep per night.

Sleep is definitely important for our health being, even if we still don’t know exactly why we sleep. It’s also interesting that a lot of people still consider monophasic sleep as a normal sleep cycle. The problem with the monophasic sleep cycle is that we get tired very fast during day and productivity is actually lower than it would be if we would take more naps during the day. With more short naps, we could probably also remember more dreams we have during the night.

Source: Polyphasicsociety