Doc.ai thread

[Anonymous No.3548355]

https://tokensale.doc.ai/

Anyone heard of this one? Decentralized healthcare AI project. Pretty cool, does it have the potential to moon? The ICO just started apparently.

[Anonymous No.3548753]

Skimmed through the white paper and website.
Decentralised machine learning is really incredibly hard. Consider that distribution of something as simple as a KNN algorithm is already a fucking herculean task and they are talking about DNNs there, which is orders of magnitude harder to do.

Also, what are they trying to solve here? The white paper is confusing distribution with parallelisation, the latter is the real bottleneck in deep learning right now (parallel communication between layers of a network).

All of this would require a huge infrastructural undertaking: distribution across GPUs, making it all work with CUDA, etc. No mention of how this would be solved in the paper. N real math presented either. They are just talking about distributed back propagation methods, not sure why exactly

Team is also lacking in talent to pull this off. Some belgian fuck, bunch of medical practitioners and some subpar engineers.

Then, just to sprinkle in some more shit, they mention the data problem too and how somehow its all gonna be magically taken care of with the memechain. Reality is, quantity of data is not gonna solve it. Shit in shit out. Data collection and distribution is a huge business, nobody's gonna share it with some random crypto fucks.

[Anonymous No.3548919]

>>3548753
Wow senpai, really that bad?

[Anonymous No.3549034]

If anyone wants in on this shit I'm approved for the presale (accredited investor) and can do a group buy

[Anonymous No.3549057]

>>3548753
shhhhhh. You're not playing the game right. This shittoken has THREE buzzwords
>Decentralized healthcare AI project
COUNT EM
THREEEEEEEEE
moon for sure

>>3549034
Is that an exception for a sale through an exception? Or is that just technically illegal?

[Anonymous No.3549307]

>>3548919
Yeah, I'm afraid so. As >>3549057 mentions though, THREE FUCKING BUZZWORDS. So, imminent moon obviously. Normies might jump in so fuck knows man, short term probably a good punt as long as you get out fast.

The belgian fuck has the right idea I give him that. Not a new idea by any means, but the right one. Moor's law is stagnating, so obviously "Lets distribute that shit". To EDGE networks no less, fuck. In reality though, parallelisation is required, two completely different things. And trust me, they are not gonna be the ones who crack this, with a shittier team such as this one.

Healthcare is also the right domain. Read up on computer/machine-aided diagnosis for example. And yes clean, labelled medical data is in massive demand since thats what it takes to train supervised machine learning models. But you ain't just gonna stumble into that. All you gonna get from decentralising data collection there is dirty shit. No matter how great the algo (unless we are talking about Bayesian DNNs), shit in -> shit out. No company involved in commercial data collection right now is gonna share this with them by using their blockchain. Why would they? Whats the incentive?

I hate that they simply state questionable assumptions as fact. This is supposed to be attractive to research groups. In what way? Distribution is is the solution to a perceived stagnation in processing power. How? Smart contracts on top of it all. For what reason? Distributed SGD is somehow attractive. Why?

Also, example apps on the site have no relation to what they are claiming to do. All this shit requires neither distribution or parallelisation and can be built with keras and pertained models in 40 lines of python code Laughable.

[Anonymous No.3549377]

>>3549307
I'm actually mainly wondering because I'm considering working for them lol.

I want to break into the crypto space with my career, but damn is it hard to find a project that frankly isn't completely fucking amateurish. :/

[Anonymous No.3549405]

>>3549307
But yeah, I gotta say all your points are pretty much dead-on. Can't disagree. Really makes you think.

[Anonymous No.3549581]

>>3549377
Nah skip this one anon.
If you are looking into commercial applications of crypto, focus on the financial shit right now, if your background allows you to that is. I'm talking risk management, transactions, crediting, etc. Fuck even properly designed and executed crypto debit cards are still unsolved.

All of that is actually feasible right now and you are gonna work on something concrete, not fucking castles in the sky. Everything else will come eventually, but the whole industry (let alone the tech) needs to mature, before any of it actually comes to be. 99% outside of finance is vapourware right now.

[Anonymous No.3549625]

>>3549581
Thanks anon. I'm a fresh college grad, Electrical Engineering and Biometrics. I thought this would be the perfect industry to get my feet wet. But it's looking increasingly like I'd have a better chance starting my own shit than trying to kick my career off in the crypto world, working for someone else.

[Anonymous No.3549641]

>>3549581
I've actually been thinking I'd love to make an actually user-friendly hardware wallet solution, due to my more physics and electrical background. The Ledger is pretty nice but still way too "strapped into the bitcoin world" to really get normies on board.

[Anonymous No.3549686]

>>3549057

Not sure what you're asking. Presale is open to U.S. accredited investors

[Anonymous No.3549835]

>>3549641
Can't help you there anon. I come from an abstract maths background and do quant crypto trading on the side, so you hardware people are inherently mysterious creatures to me lol.

All I can say is, crypto, from my point of view, is still very far removed from the collective consciousness. Once we reach pre-dotcom level startups market cap then its fair to start talking about all this other shit that might push the industry towards actually executing more novel ideas. Right now though, focus on applicative domains, shit that will move mr. normie towards converting his hard earned dollar to crypto.

[Anonymous No.3549914]

>>3549835
That's what I'm saying. Until my normie friends can hold something in their hand, push a button (or point it towards my device in MY hand or whatever) and move a slider to send a specific amount of <insert coin name here>coin quickly & easily, the appeal will not reach the masses. I bet it'd be super simple to do.

[Anonymous No.3549955]

>>3548355
Hey guys, if you want to buy and sell whitelist accounts, join us on Whitelist Trader Slack:

https://join.slack.com/t/whitelisttrader/shared_invite/MjM1NzA3MjkxNTI0LTE1MDQ0NTkzMzQtMTA2ZDBmNmZjMw

[Anonymous No.3550157]

>>3549914
Thats true and I agree, user-friendliness is fucking autistic right now... So anything in that space will help. What I'm talking about though goes a little bit beyond that. All the decentralisation and distribution talk and all that is essentially rubbish to normie ears, crypto needs a killer app, that would take it past the speculative only domain.

Obviously ease of transactions is needed. But then take gambling for example, seems like the perfect fit. Freedom of negative credit repercussions for all the degenerates out there. Potentially freedom from the long dick of the regulatory bodies (potentially being the keyword here).

Or if you wanna take it a little further, look into what Walton is all about. Disregard the coin itself, it will probably tank because they are so exposed to all the China fud.

The technology itself though makes a lot of sense, again if executed correctly. Combining RFID or any other form of tracking with a transactions chain, would allow me as a normie to quickly and easily check if that iphone I'm buying off of ebay is counterfeit or not. Tech is already there, automated tracking is already in wide use, just needs proper execution.