Tau-Chain and Agoras Official Thread: Generalized P2P Network - page 81.

FrictionlessCoin

legendary

Activity: 868

Merit: 1000

Cryptotalk.org - Get paid for every post!

Quote from: ohad on May 10, 2017, 05:48:12 PM

is a special case because machine learning algos can be formalized in logic.
eth's coordination is over the blockchain only, and that's part of their limitation. it's "code inside blockchain", not "blockchain inside code". on tau, it gives rise to reason over the network as a whole. in other words, if eth would replace the contract language to decidable one, they still cannot assert that the contract won't break the global eth network/economy.
feel free to discuss machine learning or blockchain stuff with me, here or in any other media

So if I understand your approach correctly, you want to do something like Tezos (i.e Nomic), where all rules are immutable?

That's an interesting research idea, but I'll watch the area to see how it progresses.

I don't know yet if I need that kind of P2P network, but we shall see.

ohad

hero member

Activity: 897

Merit: 1000

http://idni.org

is a special case because machine learning algos can be formalized in logic.
eth's coordination is over the blockchain only, and that's part of their limitation. it's "code inside blockchain", not "blockchain inside code". on tau, it gives rise to reason over the network as a whole. in other words, if eth would replace the contract language to decidable one, they still cannot assert that the contract won't break the global eth network/economy.
feel free to discuss machine learning or blockchain stuff with me, here or in any other media, e.g. email ([email protected]) or irc (#zennet on freenode)

FrictionlessCoin

legendary

Activity: 868

Merit: 1000

Cryptotalk.org - Get paid for every post!

Quote from: ohad on May 10, 2017, 05:42:19 PM

Quote from: FrictionlessCoin on May 10, 2017, 05:41:18 PM

In scanning your website, it appears that you are taking a GOFAI (Good Old Fashion AI) approach.

That is different from Intuition Fabric (INT) that uses a Deep Learning approach.

It is the difference between the Symbolists approach to AI versus the Connectionists approach to AI.

almost.
is the difference between learning from example using statistical methods,
to rule-based-ai. to just use logical inference.
the former is a special case of the latter, not vice versa.
further, the latter makes sense in a trustless p2p network, not the former.

Yes, I think we can agree on the differences in the approach.

I would disagree though with the former being a special case. Humans aren't built like machines, but are capable of logic inference.

I use the Ethereum network for P2P capabilities. I don't use it as part of the inference process.

Anyway, best of luck with your project!

I will be at the NYC Token Summit on May 25th, we can chat then if you are attending.

ohad

hero member

Activity: 897

Merit: 1000

http://idni.org

Quote from: FrictionlessCoin on May 10, 2017, 05:41:18 PM

In scanning your website, it appears that you are taking a GOFAI (Good Old Fashion AI) approach.

That is different from Intuition Fabric (INT) that uses a Deep Learning approach.

It is the difference between the Symbolists approach to AI versus the Connectionists approach to AI.

almost.
is the difference between learning from example using statistical methods,
to rule-based-ai. to just use logical inference.
the former is a special case of the latter, not vice versa.
further, the latter makes sense in a trustless p2p network, not the former.

FrictionlessCoin

legendary

Activity: 868

Merit: 1000

Cryptotalk.org - Get paid for every post!

In scanning your website, it appears that you are taking a GOFAI (Good Old Fashion AI) approach.

That is different from Intuition Fabric (INT) that uses a Deep Learning approach.

It is the difference between the Symbolists approach to AI versus the Connectionists approach to AI.

Anyway, do let me know when you get your whitepaper out. I definitely would be interested in reading it.

Mine is out, so you can always read it.

ohad

hero member

Activity: 897

Merit: 1000

http://idni.org

Quote from: FrictionlessCoin on May 10, 2017, 05:37:12 PM

Quote from: ohad on May 10, 2017, 04:27:08 PM

Quote from: FrictionlessCoin on May 10, 2017, 04:04:50 PM

Quote from: dmitryshech on May 10, 2017, 03:43:44 PM

Quote from: AEA on May 10, 2017, 03:25:38 PM

Quote from: ohad on May 10, 2017, 02:51:49 PM

Quote from: John_Paul on May 10, 2017, 02:31:02 PM

Quote from: ohad on May 10, 2017, 11:08:56 AM

100% scam
like 99% of projects in this bubble/pyramid era
even tezos which i thought were decent at least biz-wise, i've been notified that they copy from my writings (and even my mistakes
https://hackernoon.com/smart-contracts-turing-completeness-reality-3eb897996621
many do that, and i'll take them all to court one day, but it's the first time i saw that tezos do something like that

That is why you probably should wait as long as you can to release the whitepaper (or maybe a short version first then the full paper when it is needed.)

the plan is to finish the paper first, that'd be safe and successful to my estimation, but, if buyers will repeat this request from me due to e.g. fear of competition (i.e. if more people agree with you), i can postpone publications by putting time into code

(i'll reply later with little more detail about tezos)

I'd rather you code. It's all about the product. Most people won't even understand the whitepaper anyway.

The whitepaper will only raise the price of the agoras. Product, product and product. We move forward!

Agree. I also don't mind waiting with the white paper. This is dog eat dog environment. It's not the first time I mention Microsoft here but Billy also stole someone's ideas, look where he's today... There is a lot of hype around Tezos now http://www.cnbc.com/2017/05/05/billionaire-investor-tim-draper-backs-new-cryptocurrency.html so this is a very tough decision imo and everything should be taken into account. I guess it also depends on how far they can go with what they already copied

Just want to add a note that Intuition Fabric's whitepaper is out. It weighs in at 124 pages: http://intuitionfabric.launchrock.com/

Of course, if you are talking about something else, then my apologies.

generally speaking, some buzzword combination, no matter how popular, cannot possibly make sense. deep learning + blockchain is one of them.
i'll be glad to be proven wrong but i also have proofs to be right.
machine learning was was my main field of r&d before i began with crypto. if there was a sound way to combine the two, i'd do it first.

Well, I don't really know exactly what your plan here is, however rest assured that Intuition Fabric's (i.e. INT) approach to combining Deep Learning with Blockchain technology is quite innovative. You should read the whitepaper.

Well, clearly you won't be able to do it first since you just said that 'it cannot possibly make sense'.

But just to be clear, whatever you are planning with Agora Tokens, it definitely would be very different from what we are building with Intuition Fabric. So there's no conflict here. Just a very different approach.

indeed by "generally speaking" i referred mainly to not reading the whitepaper (which cost $8 to read which is weird but i can understand why people do so, not agree much though), and indeed there's no conflict at all as that's not what tau tries to do, as i dont see any added value in doing such a thing.
added value in combining machine learning with blockchain. not in machine learning in general. "deep" is just a shallow buzzword