EFFECT.AI : Decentralized Network For Artificial Intelligence On NEO Platform

Over the years around us, the growth in the number of Artificial Intelligence (AI) applications has skyrocketed. Some Smart services such as self-driving cars, face and voice recognition in mobile phones and image translation has been the center of attraction in our everyday life. This increase can be attributed to advances in machine learning research and the ready availability of cloud computing. There has been a great adoption of this mechanism by industries and the conception has resulted to producing a billion-dollar-economy around smart applications. Most intelligent algorithms are being developed behind the closed doors of large corporations while academic achievements are being made known to the public.

The Effect Network is a private, decentralized ecosystem which is designed to make provision for a feature complete alternative to services such as Micro tasking, ML computation etc. as well as operating fully on smart contracts deployed on a Turing-complete blockchain. This model proposes a protocol that decentralizes the global market in Artificial Intelligence, hence, lowering the barrier for entry, stimulating market growth and greatly reducing usage cost.

ARTIFICIAL INTELLIGENCE MARKET

The benefit of the AI market from decentralization is as a result of the high degree of interaction between agents. To facilitate these interactions and have a boost in collaborations, an open and decentralized ledger could be employed. The following below are the major reasons why AI development is currently inaccessible:

Data Processing
Certain tasks that traditionally require human feedback can be performed by Intelligent applications. These tasks involve processing unstructured data and finding patterns which are necessary to provide useful output. One important thing is, these applications are trained on large data sets with annotations. This makes things very simple as obtaining annotated data set is non-trivial and requires a lot of time and money.

Diverging Tasks
One problem faced with developing a complex algorithm is the need for an interaction with parts of the world outside the current niche or location e.g. a self-driving car will have to learn how to steer as well as become familiar and identify all road signs around the world. This is a knowledge system situation where classification of sign is being done by an external application. Hence, there is an increase in the amount of work needed.

Computational Cost
A very computational intensive task is the developing and training of a large AI. This process requires a technical infrastructure which is capable of handling and processing terabytes of data, doing batched processing on multiple GPUs and coordinating the results.

The three points explained above are being solved by the Effect Network and like every other decentralized application, Effect does a connection between supply and demand without an intermediary party interfering. The advantages associated includes:

• Accuracy : There is a form of exchange called the Effect Exchange which is rich in ontology of specialist AI applications. Individual applications have the ability to find each to make transactions such as to buy and to sell information. Through this exchange, users can use data sets with significantly higher complexities to train their AI algorithms.
• Accessibility : Through the direct linking of supply to demand by the Micro tasking platform, Effect Mechanical Turk which will make training AI algorithm easier, cheaper and faster. There is an enhancement for users who do not have access to a large dataset or a big network to train their own AI algorithm.
• Performance : On Effect Exchange, users can directly buy existing data sets or better still, create their own data set by creating micro-task on the Effect Mechanical Turk platform. The possible way to use datasets to train AI algorithms is by enabling users to retrieve accurate data sets quickly.
• Interoperability : To decentralize AI and achieve interoperability between individual AIs, putting the AI algorithms on the blockchain and creating a standard to which these AI algorithms comply to. The combination of multiple AI algorithms will result in emergent intelligence and powerful capabilities to which no single algorithm can achieve on its’ own.

PHASES OF EFFECT.AI

These phases are modes in which the network is deployed. This allows adaptation and the development of the network to grow together. These phases are interconnected in the network model while covering independent market sections. It is fueled by a token called AIX.

Phase 1 : Decentralized Mechanical Turk
The Effect Mechanical Turk is a decentralized, peer-to-peer marketplace for tasks which requires human intelligence. There are some familiar and similar features provided such as centralized services like Fiverr, Amazon Mechanical Turk, Crowdsource and Guru.com. This is a crowd sourcing technology that allows requesters to submit task that can be completed by human agents in exchange for compensation. These tasks are called Human Intelligence Tasks (HIT) and the providers of these tasks are called Requesters. When a task is completed by a worker, he or she is paid with a cryptographic token called AIX.

Key roles :

1. Requesters
2. Workers
3. Tasks
4. Data sets
5. Privacy

Phase 2 : Decentralized AI Exchange
The Effect Exchange is a decentralized marketplace where services of AI algorithm can be exchanged. The model works with an application owner registering on the exchange by specifying a public endpoint of his application in accordance to the technology’s data interchange format as well as usage fee for consumers. This then allows the application to be involved through smart contracts on the blockchain. In addition, the contract caller will have to transfer the required funds to the contract owner for an identification and authorization token that permits his interaction on the application.

Key roles :

1. Application Registry
2. End points

Phase 3 : Decentralized AI Algorithms
Phase 1 and Phase 2 of the Effect Network decentralized every data gathering and usage of AI algorithms. Till date, algorithms are still being run on centralized servers. The final phase of this network deals with the distribution of actual computation to make sure algorithms run globally without having to encounter failure during the process. In order to achieve this protocol, the use of machine learning algorithms having a rigid structure should be employed as well as operating on sets of weights. The advantage is that this type of algorithms is relatively easy to distribute. This will run popular deep learning frameworks such as Caffe, MXNet and Tensor flow.

ICO/TOKEN SALE DETAIL

Effect Token Distribution details goes thus:
Start Date ---------------------------- March
End Date ----------------------------- 18 Days after Start Date
Soft Cap ------------------------------ € 4,280,000
Hard Cap ----------------------------- € 14,820,000
Discount ------------------------------ 10%
Minimum invest ---------------------- € 50
Maximum invest --------------------- € 25,000
Accepted currencies ---------------- NEO, ETH
Maximum numbers of tokens generated ------- 650,000,000 tokens
Token Sale percentage ----------------- 40%

Distribution Details
40%: Public Token Sale
20%: Future Funding
16%: Platform Incentives (Galaxy Pool)
15%: Effect.AI Team
7%: Partnerships and Advisors
2%: Bounty

Funds Distribution Details

• Blockchain Platform
  20% : Technology Development and Effect Mechanical Turk Platform.
  13% : Advanced Functionality of Blockchain and Smart Contracts.
  10% : Security and Hardware

• Research
  3% : Blockchain
  2% : AI
  2% : IOT

• Business
  18%: Requesters and Workers Acquisition
  15%: Marketing
  9%: Operations, Overhead, Legal, Administration
  8%: Partnership and Talent Acquisition.

In conclusion, having an open, accessible and an affordable platform for intelligent algorithms will be an important key component to operate and develop in the future. A perfect match is an Artificial Intelligence and Decentralization. The Effect Network combines these two technologies (blockchain technology and the surge in the global AI-market opening room for more opportunities) effectively.

TEAM

Website
Whitepaper
Telegram
Twitter