The second transformation; everything about it, screams DMT, especially mushrooms.



A little backstory:

first tests playing with #deepdream #inceptionism

A visualisation of what's happening inside the mind of an artificial neural network.

In non-technical speak:

An artificial neural network can be thought of as analogous to a brain (immensely, immensely, immensely simplified. nothing like a brain really). It consists of layers of neurons and connections between neurons. Information is stored in this network as 'weights' (strengths) of connections between neurons. Low layers (i.e. closer to the input, e.g. 'eyes' store (and recognise) low level abstract features (corners, edges, orientations etc.) and higher layers store (and recognise) higher level features. This is analogous to how information is stored in the mammalian cerebral cortex (e.g. our brain).

Here a neural network has been 'trained' on thousands of images - i.e. the images have been fed into the network, and the network has 'learnt' about them (establishes weights / strengths for each neuron). (NB. This is a specific database of images fed into the network known as ImageNet image-net.org/explore )

Then when the network is fed a new unknown image (e.g. me), it tries to make sense of (i.e. recognise) this new image in context of what it already knows, i.e. what it's already been trained on.

This can be thought of as asking the network "Based on what you've seen / what you know, what do you think this is?", and is analogous to you recognising objects in clouds or ink / rorschach tests etc.

The effect is further exaggerated by encouraging the algorithm to generate an image of what it 'thinks' it is seeing, and feeding that image back into the input. Then it's asked to reevaluate, creating a positive feedback loop, reinforcing the biased misinterpretation.

This is like asking you to draw what you think you see in the clouds, and then asking you to look at your drawing and draw what you think you are seeing in your drawing etc,

That last sentence was actually not fully accurate. It would be accurate, if instead of asking you to draw what you think you saw in the clouds, we scanned your brain, looked at a particular group of neurons, reconstructed an image based on the firing patterns of those neurons, based on the in-between representational states in your brain, and gave *that* image to you to look at. Then you would try to make sense of (i.e. recognise) *that* image, and the whole process will be repeated.

We aren't actually asking the system what it thinks the image is, we're extracting the image from somewhere inside the network. From any one of the layers. Since different layers store different levels of abstraction and detail, picking different layers to generate the 'internal picture' hi-lights different features.

All based on the google research by Alexander Mordvintsev, Software Engineer, Christopher Olah, Software Engineering Intern and Mike Tyka, Software Engineer

Wow!

Amazing!

Oooh my so cool

Was it the same layer getting enhanced every time, or were you varying the behavior of the net?

Blessings
~ND

Tbh, I have not the slightest clue about 'the same layer getting enhanced'. Im still reading into all this tbh.

When you say 'were you'; apologies, I don't follow; I wasn't varying anything hehe, this wasn't me nor did I do this video.

Btw, the background info at the top was copy and pasted from vimeo's original video of this. Juist a reminder for people reading/watching this. I didn't write it. Should've put it in 'quote'.

<3

Omg we are getting so close to full AI that it's scary

Oh, woopse, I misunderstood, I thought that was you in the video. My bad.

The neural net used to make these effects has multiple layers, higher layers process higher level abstractions and more complex imagery, similar to how our brains are thought to do object recognition.



Meh, I don't buy that, I'm not sure we'll see truly strong AI until the next century, if we get there at all. There's so much more than the ability to make pretty pictures. Google's image recognition software is still rife with problems, and UAVs screw up way more than they're successful.

I'm not even sure what 'AI' means at this point? Is it the ability to perform a given task? Passing the turing test? Creativity? 'Consciousness?'

Blessings
~ND

Ahh okay. Thanks for a little bit of explanation. No worries.

After looking at some of the images on googles tech page on twitter (where people are uploading their images through this prog), I see what you are talking about and while im not sure about the video I linked; many of the images uploaded on their page are being passed 25-50 times through apparently (at least by a few people). Id like to know about the video I linked as well..

<3

What happens if you coalesce the different extracted layers and combine them? Or is that what we are seeing already?

I'm not sure I understand what you mean. Each layer processes a different aspect the input image. Layer 1 might just pick out lines, and angle orientations, which then feeds that information to the next layer, which uses those lines and angles to make shapes, layer 2 feeds into layer three which uses those shapes to make slightly more complex shapes.

By enhancing the output of a layer, you can 'see' visually, what each layer does. It doesn't make sense to talk about 'coalescing' the layers I think, since they're all already working together. Again, it's analogous to the human brain, where one task (object recognition for example) is broken down and distributed across many different regions, all of which do a unique part, but communicate with each other to create an integrated percept of vision.

Blessings
~ND

The second one is ridiculously similar to DMT. It actually gave me shivers!

Here's another neural network generated vid:

I find it interesting that this is a Google AI thing . . . if you've seen Ex Machina you'll know it was about a billionaire founder of a search engine that may as well be Google (they called it something else, Blue something) and that he created the first real AI . . . . lots of movies and shows and news about robots and AI lately. I, for one, welcome our new robot masters.

Just when I started to think that Fear and Loathing couldn't get any weirder/trippier...



Thanks for sharing Kermit!

THAT was something else! I cannot wait to see where this technology brings us

All sorts of new DeepDream content is popping up across the internet. Google has shared how to create these images ourselves here. You can view what other people are coming up with here (WARNING--some images may be NSFW).

I'd love to see what kind of weirdness folks here are capable of creating

Someone needs to put Enter the Void on this thing.

This Deep Dream software is amazing.

https://vimeo.com/132700334

This is pretty interesting:

Does anyone else think that this seems to closely resemble what our brains are possibly doing?

That was my first thought!

How is it that this "AI" can produce images that so closely resemble the altered states of consciousness brought on by many psychedelics? I sat pondering this for several hours during a trip recently and I'm no closer to any sort of reasonable answer.