Photographic Gems of Our Times From The Web Inceptionism: Going Deeper into Neural Networks

A fascinating article about neural networks. Neural Networks are used in image recognition. The idea is, basically, to build a system mimicking human vision and to let the machine train itself to recognise objects. But because we let the machine train itself, the resulting model is not something that can be described with a simple rule set, and scientists have difficulties grasping how the resulting model works.

The article presents what happens when we tried to tweak the image so that the recognising process becomes more visible. For example, here, the low level networks which are used to recognise lines and shapes:


A different example where the higher level networks are extracted. Here we see what the networks considers to be the "essence" of various subjects:


As humans we are gifted with the ability to process images into something we understand, but also to immediately decide whether an image is something we like to watch because we find it aesthetic or whether it is poor, disharmonious or unbalanced, but we don't know why. In essence, we are gifted with the capacity to "read" pictures, be we don't know how we do it. This kind of research allows you to get glimpses into these processes, to read the reading.

This concept of "Inceptionism" is new to me. I am so impressed by the stacked layers which each have a distinct realm of work to do and can be somehow trained to get experience.

It's impressive how birds are found in clouds and even fused dog and fish creatures.

I wonder whether analagous processes are involved In the development of the internal self-consistency of various religious doctrine, each so convincing to adherents but actually each often mutually exclusive in their core beliefs.

If one teaches one set of doctrine, and there's no outside influence to challenge this, then we will "discover" that life does indeed follow the rules we were taught, (but which were actually invented and evolved to support that doctrine), as they are internally self-consistent!

I find the studies with picture recognition very interesting and I sense it is going to reveal a lot about how we function in more than just interpretation of images or creating new art!

Asher

Asher Kelman said:

I wonder whether analagous processes are involved In the development of the internal self-consistency of various religious doctrine, each so convincing to adherents but actually each often mutually exclusive in their core beliefs.

Click to expand...

Maybe, since our thought processes are all based on neural networks tuned to recognise patterns. But I have no idea about how to study the formation of religious doctrines scientifically, so I would rather pass on that subject.

Jerome Marot said:

Maybe, since our thought processes are all based on neural networks tuned to recognise patterns. But I have no idea about how to study the formation of religious doctrines scientifically, so I would rather pass on that subject.

Click to expand...

When the computer neural networks are inderstood, I suspect we could gain more insight into ourselves! After all, the geometries to store, organize and stratify data and then build assumptions based on that, seems to have much in common. Moreover, as the neurobiologist gain definitive anatomical constructs of our own neural organization, our modeling with computer based neural networks could be fashioned close to what we ourselves are built with. So I see a natural iterative development of insight. There's a huge potential here and it will become progressively "unbiased" except in the closeness we approach what neurobiologists discover in parallel research.

Asher