Intuition Fabric - Decentralized AI on Ethereum

IntuitMachineIntuitMachine Posts: 2Member
edited April 12 in Smart Contracts and Dapps


Deep Learning is an advanced form of Artificial Intelligence. A blockchain is a distributed transparent consensus based peer-to-peer network that has shown to be extremely resilient to adversarial attacks. This paper explores the "Intuition Fabric" a fusion of Deep Learning technology on to a blockchain. The goal of this platform is to build a system that leads to the democratization of AI.

Details can be found in the 124 page Whitepaper:

http://intuitionfabric.launchrock.com/

alternative whitepaper link: https://gum.co/wxLP

Use "avalonbeta" as code to access the whitepaper.

Intuition Fabric is developed by Intuition Machine Inc. http://www.intuitionmachine.com

All inquiries send to info@intuitionmachine.com

Blog: https://medium.com/intuitionmachine
Facebook: https://www.facebook.com/groups/deeplearningpatterns/
LinkedIn: https://www.linkedin.com/groups/8584076
Newsletter: https://www.getrevue.co/profile/intuitionmachine
Twitter: https://twitter.com/IntuitMachine


Conditions

You will need to do one of the following to be eligible:

(1) Be a Follower on Twitter +1
(2) Follow on Medium + 2
(3) Sign up at http://intuitionfabric.launchrock.com/ +2
(4) Download the whitepaper +5
(5) Be a member on Facebook +10
(6) Be a member on LinkedIn +20

We will be giving out points for each level.

Evidence of above participation will be requested prior to token distribution. Any, evidence of double registration or fake profiles will be grounds for disqualification. Note, only (1) and (2) are automatic.

We will ask this information at the time of distribution. A form will be provided that you can fill in. Your points will be proportional to the amount of tokens that you will receive. The exact number will be determined at the time of distribution.

Developers Wanted

Interested developers please PM to join the team!

Skills required: Go, Javascript, Python, Java, Ethererum, Tensorflow, PyTorch.

Added plus: Deep Learning, Reinforcement Learning, Game Theory, NLP, Non-linear Statistical Mechanics, Quantum Tensor Networks

Bounties

0.5 ETH (Ethereum) for acceptable translations of the whitepaper.

1 ETH for persuasive "sales copy" pitching Intuition Fabric.

Contest

1 ETH for a winning infographic to be created that explains Intuition Fabric (INT).

0.5 ETH for a winning logo for INT

Miscellaneous

Here is an inspirational video about Deep Learning:



and a trailer for a movie about Intuition:



TEDx talk on Intuition:



TED talk on Intuitive AI:



Denis Hassabis on latest developments



"Human Ingenuity augmented with AI will unlock our true potential"

Comments

  • o0ragman0oo0ragman0o Posts: 1,236Member, Moderator mod
    Ok, I've just downloaded your 124page (!) white paper and am still pretty clueless as to what you're actually doing, why you need blockchain to do it, why it requires 'market incentive mechanisms, but most of all why I need to have Facebook and LinkedIn profiles (of which I'm deeply opposed to) in this field of blockchain? Also why would someone want to pay $8 for your whitepaper when there is no summary or anything about what it contains? Had you not provided the promo code, I certainly wouldn't have look at it. I simply don't have time to read it.

    It's only that I've had a life long interest in AI (writting chatbots in the '80 ANN's in the '90s, Udacity's first sefldriving car course more recently) that I've looked at it at all.

    It comes across as a huge jumble of pointless hyp'ed op-ed pieces about the current AI technologies available
    In short, your presentation is very loose and you need to be far more succinct about what it is you are actually wanting to build on blockchain.

    From my brief scanning of the WP, It appears that:
    1. you want to store ANN training data on IPFS.
    2. want to incentivise a community of developers (?)
    3. integrate bigdata
    4. ???

  • IntuitMachineIntuitMachine Posts: 2Member
    Apologies for a delayed response.

    I guess in today's distracted/multi-tasking environment, I can't expect people to read and dissect the paper.

    Well, you can't criticize it a jumble of pointless sections, but they are all there for a purpose to build up the idea.

    To summarize:

    1. Deep Learning models are stored on IPFS to be executed by a DL VM.
    2. Training data can optionally be stored on IPFS so models can be recreated
    3. Users can use the models on IPFS to run locally or use a cloud provider
    4. User's data used for inference are therefore private.
    5. Trainers contribute new models by deriving from existing models.
    6. Trainers are compensated for their models using Tokens.
    7. Users access the network using Tokens.
    8. Bounties will be available to encourage developers to create the latest SOTA (state-of-the-art) models. Compute resources will also be made available to developers.
    9. There are other roles such as Curators and Auditors that participate to improve the network.

    The idea is to build a democratized Deep Learning system that anyone with Tokens have access to perform their daily activities while preserving their privacy. So for example:

    (1) Automatically reading email and replying.
    (2) Automatically categorizing own personal data.
    (3) Advanced searching of confidential data.
    (4) Conversion of confidential unstructured data into knowledge bases.
    (5) Reducing information overload.
    (6) Running prediction algorithms that watch markets.

    The list can be very big. The main idea is to allow 'frictionless' access to these capabilities while still preserving one's own privacy.
Sign In or Register to comment.