Was thinking about specializing in surgery this month because of clinical rotation, but man, every time I look at one of your videos, my love for engineering and applied sciences just keeps growing, thanks for your effort and quality
The best video for me that interrelates the properties of DNA with AI, I already knew all this, but seeing the two concepts together...(In the end any system that uses logic can emulate intelligence)
It should be added that I hope that in the next one it will be how DNA does backpropagation, the algorithms that allow, just like AIs in training, to readjust their weights...
Amazing, can't wait for the series! Just leaves me to wonder, could it be possible to solve optimization problems by using evolutionairy pressure? Maybe by using dna or proteins (or even on bacterium scale) to produce a desired output and then apply evolutionairy pressure to optimise the calculation/production process? A bit like machine learning but using evolution as an optimisation mechanism. If anyone would be able to come up with a way to do this, it would definitely be you :D
ooh so interesting! ive been thinking about using nanotechnology for neural networks for years! i would like them to be able to move around and edit themselves intelligently, scouting for new architectures to reconfigure itself into
Is it practical to approximate a sigmoid with polynomials? I would have assumed we use some sort of saturating reaction that is inherently present in the inner workings of a cell to get a sigmoid approximation. Like you said about the AND gate (starting at about 8:50): it doesn't matter how much we add of the seconds part, the reaction is saturated. The result can be expressed as f(x) = 0 for x < 0; x for 0 < x < B; B for B < x < inf. (x is just a renaming of the amount of A present in the cell)
It is reversible, easily in fact. But the precise control we have with solid wires would be extremely difficult to replicate with DNA (or any chemical). Even running the simplest program would be an engineering nightmare. But if we ever could make this work with precise control it’s an interesting concept for sure. A more practical and achievable type of coding may be achieved by something like the “from silicon to cells video” But that still has a lot of problems. For example speed will always be a barrier (it’s the reason we have electrical signals in our nervous system and not chemical signals), and it will also almost certainly be less energetically efficient. Not to mention all the engineering hurdles that would need to be overcome to may it work in the first place.
![The moment we stopped understanding AI [AlexNet]](http://i.ytimg.com/vi/UZDiGooFs54/mqdefault.jpg)
![The moment we stopped understanding AI [AlexNet]](/img/tr.png)
Can't wait to see what you can do in a living cell, this is nanotechnology.
How do you have 10.5k subscribers with 0 videos?
"There's plenty of room at the bottom."
And that’s why my channel is named “NanoRooms”. There’s plenty of room at the bottom (nano scale).
*distant* "ayyeeeoo-"
Was thinking about specializing in surgery this month because of clinical rotation, but man, every time I look at one of your videos, my love for engineering and applied sciences just keeps growing, thanks for your effort and quality
holy hell, this is pretty poggers. there's a lot of cool stuffs we could do with this tech
DNA computing was one of the most mind-blowing things that I learned about in my undergrad. Super inspiring stuff and I'm glad you covered it :)
Glad to see you again! Great video as usual.
Thanks!
The best video for me that interrelates the properties of DNA with AI, I already knew all this, but seeing the two concepts together...(In the end any system that uses logic can emulate intelligence)
Will be waiting for the next videos
This is beautiful, can't wait for the rest of the videos!
Really interesting video. And I find it really funny that this video was the first to make the math of neural networks finally click for me.
This is so wonderful! Your videos are just amazing and deserve more praise.
It should be added that I hope that in the next one it will be how DNA does backpropagation, the algorithms that allow, just like AIs in training, to readjust their weights...
Amazing, looking forward to your next video in this series and any more videos relating to synthetic biomolecular computation.
Such an under rated channel! Can't wait to see the two new videos!
This is fire, idk how you've got only that few followers
This channel has gone on such a journey from SoME2 to SoME3 🎉
Amazing, looking forward to your next video！
This channel is so fucking cool.
I Hope this blows up soom
Amazing, can't wait for the series!
Just leaves me to wonder, could it be possible to solve optimization problems by using evolutionairy pressure? Maybe by using dna or proteins (or even on bacterium scale) to produce a desired output and then apply evolutionairy pressure to optimise the calculation/production process? A bit like machine learning but using evolution as an optimisation mechanism. If anyone would be able to come up with a way to do this, it would definitely be you :D
matrices are surely important. matrix-ception?
ooh so interesting!
ive been thinking about using nanotechnology for neural networks for years! i would like them to be able to move around and edit themselves intelligently, scouting for new architectures to reconfigure itself into
soon computers can process chemical bonding, modify DNA instructions, create living organisms and computer monitors will be like aquariums.
Can you do a booklist video so that we can go in depth
Genetically inherited memories?
Are you going to make a dragon synthesis tutorial? XD
Is it practical to approximate a sigmoid with polynomials? I would have assumed we use some sort of saturating reaction that is inherently present in the inner workings of a cell to get a sigmoid approximation. Like you said about the AND gate (starting at about 8:50): it doesn't matter how much we add of the seconds part, the reaction is saturated. The result can be expressed as f(x) = 0 for x < 0; x for 0 < x < B; B for B < x < inf. (x is just a renaming of the amount of A present in the cell)
All of this will be answered in the next episodes (or you can read the papers in the description)
Funny how DNA spent millions of years developing NN out of cells and we’re over here trying to get it to do the work itself
Is this process reversible/Can you use this process for more than one computation?
It is reversible, easily in fact. But the precise control we have with solid wires would be extremely difficult to replicate with DNA (or any chemical).
Even running the simplest program would be an engineering nightmare.
But if we ever could make this work with precise control it’s an interesting concept for sure.
A more practical and achievable type of coding may be achieved by something like the “from silicon to cells video”
But that still has a lot of problems. For example speed will always be a barrier (it’s the reason we have electrical signals in our nervous system and not chemical signals), and it will also almost certainly be less energetically efficient.
Not to mention all the engineering hurdles that would need to be overcome to may it work in the first place.
@@ThatSpazamataz makes sense thanks for explaining
This started looking promising, but it's falling apart:
"A cat is an animal, and a dog is an animal..." (Yes...) "Therefore a dog is a cat." (No.)
3:40
Yup it's getting weird
ai bacteria?
calm down god