Building a fly brain in a computer

#1

Post by breck » Mon, Oct 29, 2018, 5:08 PM UTC

Building a fly brain in a computer
cifar.ca

Re: Building a fly brain in a computer

#2

Post by youwishboye » Tue, Oct 30, 2018, 7:33 AM UTC

mother of misleading titles

Re: Building a fly brain in a computer

#3

Post by bastih » Tue, Oct 30, 2018, 8:25 AM UTC

It's interesting to see biological system and implementation next to each other. Are there any similar visualizations for other ML-implementations?

Re: Building a fly brain in a computer

#4

Post by greenstonekid » Tue, Oct 30, 2018, 9:00 AM UTC

I wonder what the computational difference between the fruit flies visual system and the computer system used to run the model is?

Re: Building a fly brain in a computer

#5

Post by jobigoud » Tue, Oct 30, 2018, 9:05 AM UTC

Can it feel pain? I know pain is kind of tricky to evaluate in biological insects as well but that's a discussion we will have to have at some point.

Re: Building a fly brain in a computer

#6

Post by speps » Tue, Oct 30, 2018, 9:05 AM UTC

bastih wrote: post #3

It's interesting to see biological system and implementation next to each other. Are there any similar visualizations for other ML-implementations?

This is a similar project: http://openworm.org/

Re: Building a fly brain in a computer

#7

Post by Hei1Fuya » Tue, Oct 30, 2018, 9:19 AM UTC

jobigoud wrote: post #5

Can it feel pain? I know pain is kind of tricky to evaluate in biological insects as well but that's a discussion we will have to have at some point.

The headline is misleading, they built a neural network based on fly visual neurology, not a whole fly brain.

And even if it were a full brain, effective societal consensus seems to be that insects don't have a right to humane treatment.

Re: Building a fly brain in a computer

#8

Post by kpil » Tue, Oct 30, 2018, 9:20 AM UTC

jobigoud wrote: post #5

Can it feel pain? I know pain is kind of tricky to evaluate in biological insects as well but that's a discussion we will have to have at some point.

I wouldn't worry about that until much much larger animal models. Besides, it's just the visual system.

I don't know where there's a moral limit. Several millions of neurons? A billion?

Re: Building a fly brain in a computer

#9

Post by bwbw223 » Tue, Oct 30, 2018, 11:10 AM UTC

Sounds buggy...

Re: Building a fly brain in a computer

#10

Post by viraptor » Tue, Oct 30, 2018, 11:50 AM UTC

jobigoud wrote: post #5

Can it feel pain? I know pain is kind of tricky to evaluate in biological insects as well but that's a discussion we will have to have at some point.

I don't think simulated brains can feel pain any more than a float variable "current_pain_level" set to 100 can "feel pain".