VisionX: A Cross-Industry Collaboration Platform for Blockchain AI Solutions
1. AI and Industry into the Future
Artificial Intelligence (AI) is already everywhere around us. AI has applications in areas such as medical diagnosis, remote sensing, robot control, and voice and facial recognition, to name a few. However, widespread application of AI is still in its infancy and many industrial sectors remain under-served by AI solutions, leaving large untapped markets for AI innovations.
Investment in the broader AI sector is on the increase as governments form strategic policies to fund research and development, recognising AI as a key strategy area for industrial and national advancement [1,2]. For example, the Chinese government is investing a minimum of $7 billion up to 2030 , by which time it has been estimated that AI will have increased global GDP by 14%, equivalent to $15.7 trillion .
Thanks to AI, data is being created, analyzed, and used in new ways not previously imagined by consumers, firms and governments 
Clearly, AI is an expanding area that is here to stay, and will make a big impact on how our societies will develop and become more efficient.
2. What Problems Does VisionX Address?
Barriers for companies looking to implement AI solutions include a lack of expertise and the cost of developing a solution in-house. Bringing in an external AI consultancy to provide a solution is also problematic, as a single company often cannot provide the AI consultancy with sufficient data to produce a useful AI solution.
Manufacturing processes such as visual inspection, predictive maintenance and object pick-and-place are repetitive, inefficient and prone to human error. These particular inefficiencies are found across many industries. Using visual inspection as an example, the electronics, textile, furniture, steel, mining, aviation and automobile industries must carry out visual inspections on their products and/or their equipment to identify defects. In China alone, visual inspection employs millions of people and costs $350 billion a year.
VisionX has an initial objective to deploy an AI solution for the visual inspection process, before expanding to develop and deploy AI solutions for other industrial applications, such as predictive maintenance, data curation and robotic control. The focus of this review is on visual inspection, as this is the first solution to be deployed by VisionX, and this solution serves as an intuitive example of how VisionX uses AI and blockchain technology to add value and improve the efficiency of industrial processes.
VisionX has identified practical issues for which there are industrial-scale incentives to invest in solutions.
3. The VisionX Cross-Industry Visual Inspection AI Solution
VisionX provides an incentivised, cross-industry, blockchain-based platform that permits collaboration between companies towards development and monetization of AI solutions . That sounds like a lofty goal, which indeed it is, but how does VisionX propose to achieve so much? The keyword is collaboration.
To overcome the issue of insufficient data, companies across multiple industry sectors must pool their data. In the case of visual defects, the main dataset consists of images of defects, such as dents, scratches, holes or other irregularities on surfaces, and images of the same surfaces without defects.
VisionX already has access to a database of more than a million images, including 129,000 images of defects across several industry sectors. VisionX utilises a patent-pending algorithm, called Dataonomy, to classify, group and identify the relationships between images. This allows the identification of the images within the broader dataset that are useful for a specific purpose (for example, images of steel surfaces, when inspecting steel beams).
To simplify, in order for the system to accurately identify, let’s say a scratch on steel, it must first learn how to identify a scratch on the correct surface, which it does by learning from the pre-existing annotated imagery. By pooling their images, companies can collaborate to expand and improve the solution, which benefits all collaborators. In this way, as time goes by, the datasets get larger and the solutions become more accurate. AI solutions can also be further tailored for a company’s specific requirements. Using the visual inspection solution of VisionX, accuracy can be improved from 60–70% to 95–99.9%, time to market can be reduced from months to days, at a saving of 30% or more compared to traditional solutions.
For visual inspection solutions, VisionX can supply hardware, including a small number of GPU servers, a camera and a robotic arm, and will have access to the decentralized AI computational power of DeepBrain Chain.
As well as greater accuracy and lower cost in the production line, VisionX employs a token model to provide further incentives for companies and developers to contribute their data, models, algorithms etc. Using smart contracts, with transactions recorded on the blockchain, contributors will also receive a share of payments when their contribution is used in creating further AI solutions. These incentives come in the form of the VNX token, and add a further revenue stream for companies as they naturally generate data during their production processes. Utility demand for the VNX token therefore comes from companies wishing to use the AI solutions developed by VisionX and its ecosystem of companies, contributors and developers.
The same broader model used for visual inspection will be extended to other cross-industry AI applications, e.g. data curation and predictive maintenance. Blockchain technology is used to record all data contributions, token incentives and future payments, in a transparent, certifiable and decentralized manner visible to the whole community. VisionX is a project that does have a true requirement for blockchain technology, which is deployed sensibly.
4. Token Economics
The VNX token is initially an ERC-20 token, and will migrate to the DeepBrain Chain (DBC) blockchain in early 2019. The timing and details of the migration have not been announced at the time of writing.
There is a total supply of 10 billion VNX, with 30%, 3 billion VNX, currently in circulation. The current circulating supply consists of 2.5 billion VNX sold to institutional investors and 0.5 billion VNX that were allocated for community development and airdrops.
35% of the total supply, 3.5 billion VNX, is allocated for “Mining”, which refers to the incentives given to contributors of data and services. This allocation is released steadily at a rate of 0.25 billion VNX per year, halving every five years.
A further 10% of the total supply, 1 billion VNX, is allocated to DeepBrain Chain Inc. and early investors, with 20% released after two years, and 5% released per month thereafter. The VisionX Foundation and Ecosystem is allocated 15%, 1.5 billion VNX, 10% of which is unlocked in the first quarter after the project is launched, with 10% per quarter thereafter. Finally, the VisionX team is allocated 10%, 1 billion VNX, 20% of which is unlocked after a period of 2 years, with 5% unlocked per month thereafter.
Significantly, VisionX will conduct a program of buyback and burn of VNX tokens, limited to a total maximum of 50% of the circulating supply.
The VNX team have taken care to see that tokens are released slowly into circulation. It is encouraging to see that the allocations for both the team and early investors are locked up for two years, and that the “Mining” allocation will provide incentives without flooding the circulating supply.
5. Partnerships and Collaborations
Despite being a fledgling blockchain AI project, VisionX has already made an impressive and broad range of partnerships and collaborations, across industry and academia .
VisionX, along with DeepBrain Chain, has signed a cooperation agreement with Japan SBF Consulting Co. Ltd, and has partnered with Ebiztie . The advantage of these associations is to promote the application of all parties in new international markets, and to improve connectivity between AI companies and small to medium manufacturing companies looking to harness the power of AI.
VisionX has established an agreement with Huaibei Normal University to apply AI solutions and Big Data analysis, alongside the AI computational capacity of DeepBrain Chain, to improve the analysis of breast cancer treatments.
VisionX also has a connection to the Japanese provider of batteries to Tesla, who use the defect detection technology invented by VisionX Chief Technical Officer Dr Haisong Gu. VisionX is developing an AI battery defect detection solution to replace and improve on the accuracy of the existing defect detection system.
The most impressive collaboration so far involving VisionX is with the state-owned conglomerate China Coal and Science Engineering Group Co. Ltd. that operates more than 5,000 coal mines. VisionX will implement tailored visual inspection solutions for aspects of the mining processes. It is anticipated that manual visual inspection will be replaced by VisionX AI solutions over the next three years. Beyond visual inspection, VisionX will also develop deep-learning solutions for abnormality and maintenance prediction in coal belts and underground gas networks, adding value by increasing safety and improving operational efficiencies.
The nature of the collaborations entered into by VisionX gives an idea of the diversity of the possible AI solutions. The partnerships show the ambition and connections of the team, and provides an insight into the state of readiness of the visual inspection AI solution. If VisionX can deliver the solutions to a customer as large as the China Coal conglomerate, the demand for VNX will be significant and the future will be bright for VisionX.
VisionX is currently (Q4 2018) deploying the initial visual inspection solution to pilot customers and integrating feedback to improve the product and add to the dataset. The hardware elements of the solution are currently under design.
Looking forward into 2019, Q1 targets are to establish links to customers in multiple industries who seek the visual inspection AI solution. Q1 2019 sees development of software and hardware for predictive maintenance and robotic pick and place solutions. Launch of Testnet and commencement of the Mining of VNX from the incentive pool is also pencilled in for Q1.
The remainder of 2019 targets launch of Mainnet, connecting companies across industries, expansion of the incentive-driven data/model market places, and addition of new solutions and datasets across new industries and specific applications.
Overall, the roadmap is ambitious while remaining relatively sensible. However, the roadmap will need to be fleshed out with more detail and updates in early 2019 as the project develops.
7. The VisionX Team
The VisionX team is well endowed with AI expertise. In particular, Dr Dongyan Wang, Chief AI Officer, Dr Harry Gu, Chief Technology Officer, and Hua Zhou, Principle AI Engineer, have a strong mix of impressive academic and industry AI experience. The remaining team members have a strong mix of experience in the areas in which VisionX intends to expand (deep learning, computer vision, image manipulation, human motion capture and marketing). There is a noticeable lack of blockchain expertise within the core team, but this is explained by the connection between VisionX and DeepBrain Chain (see below for further details).
8. The Connection Between VisionX and DeepBrain Chain: DECO
At the core of DeepBrain Chain (DBC) is a blockchain-based decentralized AI computation platform. Rather than buying, maintaining and having to replace expensive hardware, companies and researchers who require access to GPUs for AI computation can save up to 70% by accessing the distributed network of GPUs and high-speed storage devices of the DBC network. The savings come from the incentivised reward pool, which pays most of the fee received by GPU miners.
DeepBrain Chain will launch and expand its AI computation Mainnet in early 2019, and will also provide a marketplace for AI data, models, algorithms etc. DeepBrain Chain has also established a blockchain AI ecosystem, called DECO.
DECO consists of the DECO foundation, DeepBrain Chain as the AI computation platform, DeepToken Exchange as an AI-focused digital asset exchange, and a range of AI projects incubated by DECO. DECO will take promising AI projects under its wing, and apply the know-how of DBC to assist the implementation of blockchain technology and token economics within the operating model of those AI projects. VisionX is the first of these DECO-incubated projects, and will migrate to the DBC blockchain in early 2019.
DECO brings VisionX into an ecosystem where it may use the AI computation platform of DeepBrain Chain, and use and provide services to the other emerging DECO-incubated projects, bringing mutually beneficial value to all projects, whose tokens may be bought and sold on the DeepToken Exchange. The DeepToken Exchange uses a token burn model to return 80% of its profits to the DECO system, and to allow holders of the exchange token, DPT, to receive rewards when voting on new AI projects to be listed.
- The problems VisionX sets out to solve are in an expanding cross-industry environment, that is currently seeing industrial-scale investment towards improving efficiency.
- The project has a clear and realistic vision of how it can implement AI solutions to these real-world problems, and has developed an incentive-driven economic model that is likely to attract both large industry partners and a community of developers.
- The collaborations already in place are very promising, and an existing, functioning AI visual inspection solution is a huge advantage.
- VisionX is situated in a supportive AI ecosystem, to which it is likely to bring significant value in the coming years.
- VisionX has ambitious yet achievable plans to see significant adoption in the medium-term. If the collaborations keep coming as they have so far, the future looks very bright.
1. As China Marches Forward on A.I., the White House Is Silent. https://www.nytimes.com/2018/02/12/technology/china-trump-artificial-intelligence.html
2. Furman, J. Is This Time Different? The Opportunities and Challenges of Artificial Intelligence. Remarks at AI Now: The Social and Economic Implications of Artificial Intelligence Technologies in the Near Term. 2018. https://obamawhitehouse.archives.gov/sites/default/files/page/files/20160707_cea_ai_furman.pdf
3. AI is the new space race. Here’s what the biggest countries are doing.
4. AI to drive GDP gains of $15.7 trillion with productivity, personalisation improvements. https://press.pwc.com/News-releases/ai-to-drive-gdp-gains-of--15.7-trillion-with-productivity--personalisation-improvements/s/3cc702e4-9cac-4a17-85b9-71769fba82a6
5. VisionX: X-industry Collaboration Platform for High Performance AI Solutions. VisionX whitepaper. https://www.visionx.org/wp-content/uploads/2018/10/VisionXwhitepaper-EngV2.pdf
6. VisionX Progress Report #1. https://medium.com/@VisionX/progress-report-1-b634f3316b71