DxChain: use block chain to drive artificial intelligence

Artificial intelligence requires computing power, algorithms and data. All three are indispensable. Even if there are top experts with top algorithms, data and computational power cannot be bypassed.

For most companies, high-quality data is hard to come by. Most of the data is concentrated in the hands of big companies. Data is the core resource of the future. Big companies don't share them easily with small and medium-sized companies. For example, Google maps have user navigation data, facebook has user portrait and behavior data, and amazon has user shopping behavior data, which are the basis of their business model. They are decades of accumulation, which will only strengthen their potential energy. Small and medium-sized companies, can only hope foreign sigh.

Not only that, but ai also requires a lot of computing power and storage, which is a huge expense for start-ups.

DxChain and artificial intelligence

It's almost impossible for a startup or a project to get high-quality data in the current Internet environment. But blockchain offers an opportunity.

Blockchains can first return data to personal ownership. Users control their own data, and at the same time realize users' privacy protection by means of zero-knowledge proof, differential privacy, cryptography, etc., so as to avoid the worries of users' data contribution. At the same time, these data users can also gain profits through sales, which also encourages users to share and contribute their own high-quality data more actively.

It can be said that blockchain will bring an explosion of high-quality data. This is an opportunity for small and medium-sized projects and companies.

Through the block chain technology, DxChain can build a data trading market for both parties with data demands, so that enterprises with demand for high-quality data can obtain data more efficiently at lower cost, and users don't need to worry about their privacy leakage, and they can also get value feedback, which is a win-win situation. In traditional Internet platforms, data is often monopolized by large platforms, and users cannot be rewarded for data contribution.

In addition to the data, there is also computing power and storage. It is also a high threshold for small and medium-size companies with limited financial resources. Is it possible to reduce computing power and storage costs, and give smes a chance to compete? This is an important problem to be solved by DxChain.

DxChain tries to make computing power and storage sharing a reality through block chain technology. This will reduce computing power and storage costs, and make big data and machine learning more focused on algorithm improvement, model iteration and the development of artificial intelligence.

In summary, the combination of artificial intelligence and block chain is a perfect match. Both are the future of technology, and a combination of the two could bring each other a faster rate of evolution. Blockchain has the opportunity to accelerate ai.

DxChain tries to adapt to this trend and provides computing and storage services for big data analysis and machine learning through block chain. It also helps small and medium-sized enterprises to obtain data at lower cost and higher efficiency through data trading. This will be a great addition to the development of artificial intelligence.

Key words of DxChain

It's for my use

One thing that impressed blue fox's notes deeply was that it was good at using the advantages of "others" to build its own block chain technology, develop its own innovation, and ultimately serve the development of big data analysis and machine learning.

This is easier said than done. First of all, we need to be thorough, so that we can use it. The architecture of DxChain draws lessons from IPFS, HadoopHDFS, GFS, IoTeX, IOTA, Plasma, TrueBit, morpheo, Golem, etc., and promotes its advantages and disadvantages. For example, the layering of the chain on the chain is also partly based on the chain design of IoTex, which enables the main chain and side chain to undertake different tasks.

For example, DxChain introduced Hadoop, making it the computing engine of DxChain after decentralized design. Meanwhile, it also has its own consensus innovation, such as verification game and Provable Data Computation. DxChain's Proof of Spacetime consensus mechanism also evolved from "data ownership Proof" to make it more suitable for decentralized networks.

Chain architecture

In terms of structure, DxChain adopts the structure of chain. It needs to provide decentralized big data analysis and machine learning calculation on the basis of decentralized data storage. To achieve this goal, DxChain adopts chain structure to manage the main chain, storage chain and computing side chain.

Such an architecture is designed to facilitate the cross-chain operation of information, data, and assets, to meet speed requirements, and to achieve scalability.

DxChain is a two-layer block chain, including the main chain and side chain. The main chain stores account books and asset information, including status, transactions, intelligent contracts, etc. It stores less information. Side chains carry data storage and computing tasks.

DxChain USES an account based model to store transactions and asset information, including account status, cross-account transactions and receipts. It has regular and contract accounts. DxChain is compatible with the data structure of itaifang and consists of blocks linked by hash. The data is stored in the full nodes of the network.

The Data Side Chain is established on the P2P distributed file storage system to store non-asset information. Computing Side Chain is mainly used to complete Computing tasks, which are based on real business requirements. The cell can read the data from the data side chain and write the results to the data side chain.

When the task is complete, the final state is stored on the main chain through a smart contract. Intermediate status or task-level transaction information is stored in the side chain. The data side chain and the computing side chain realize the interoperation through the chain micro service, and the data and messages can be intercommunicated. Side chains communicate with the main chain through smart contracts.

In this way, the main chain and the side chain perform different tasks, which can be communicated through intelligent contracts or micro-services, but also have their own independence. Another advantage is that even if the side chain is in trouble, the main chain will not be affected.

The link architecture is designed to help achieve its goal of providing computing and storage services for big data and machine learning, unlike bitcoin, which is primarily used for financial transactions. The core goal of the design side chain architecture is to achieve high efficiency, scalability, and meet specific business scenario requirements.

The main chain of DxChain is maintained at a low cost. The side chain achieves high efficiency in computing and data storage, and the communication between the main chain and the side chain is realized through intelligent contracts, finally forming an overall service architecture. To achieve decentralization, storage side chains and computing side chains have their own common algorithms. This is also mentioned below.

In addition, asset transfer between the main chain and the side chain can be supported by writing the effective transaction of the side chain into the main chain. The side chain and the main chain use the same token, and the side chain can also define its own token.

Decentralized computing services

Current artificial intelligence needs a lot of computational power to support, but computational power means a lot of capital investment. To accomplish large-scale computing tasks through centralization, not to mention small and medium-sized enterprises, even the giants, are also under pressure.

DxChain provides computing power services for big data and machine learning. Its way of solving this problem is through decentralization. One is to share the spare computational power, and the other is that computational power can also be used efficiently for specific tasks.

The computational power of DxChain is different from that of bitcoin. It is not only for network security, but also for actual business needs. DxChain is not to provide a digital currency, but a decentralized computing environment.

In order to verify the correctness of the calculation, a consensus mechanism is needed to ensure it. DxChain proposed two consensus algorithms. The first is the Verification game, and the second is Provable Data Computation.

The verification game is used to verify the correctness of the calculation process. It can be proved that the data calculation can find a correct answer with a small probability of being attacked from a group of untrusted nodes.

The core roles of verification games are solvers, challengers, and judges. The solver is the miner, who completes the calculation and provides the solution. The challenger disagrees with the solver's solution. The judge provided the correct calculations.

The verification game does not care about the reputation of the player or any trusted party in the system. It has a penalty mechanism that requires both the solver and the challenger to carry out the task. Any wrong participant will lose his deposit. This will eliminate the untrusted players.

It can be proved that in data calculation, a computing task broadcasts through the network, N nodes perform tasks, and the first M nodes of the same answer are valid answers.

DxChain also integrates Hadoop for decentralized calculation. The core components of Hadoop are job tracker, task tracker, and work nodes in map-reduce. With reference to Hadoop, DxChain designed the roles of d-jobtracker and d-tasktracker. In the decentralized system of DxChain, the role requires a deposit, and the miner can honestly perform the task and get paid, otherwise the deposit will be lost.

Map-reduce is a centralized system. Its job tracker manages cluster resources and task scheduling. The task tracker manages tasks in the node while communicating with the job tracker. DxChain is a decentralized system, and it is difficult to maintain real-time communication between two nodes in distributed network. In DxChain, there is no need to check the status of task nodes, and one or more nodes are offline, which will not affect the final results. In the Hadoop system, it will know the node activity through the heartbeat. If the node has problems, the job tracker needs to redistribute tasks to the new node.

When the node completes the calculation, the job tracker sends the result to the computing side chain to verify the game or the verifiable data calculation. Calculate side chains to hold work assignments and results. Any miner node can receive a task.

Decentralized data storage service

DxChain provides a decentralized computing environment and a decentralized storage network. It stores the calculated results and calculated intermediate states as files.

The data side chain is built on p2p distributed storage networks, such as IPFS and Swarm. The data side chain is the incentive layer, which does not need to be used for data storage. Data and files are divided into small pieces and stored in p2p networks.

At the same time, the metadata and hash of the small block are stored in the side chain using a merpatricia Tree structure similar to that of the Merkle Tree, representing the file state. In order to interlink data, DxChain designed the uniform resource identifier across the chain for the file.

Between the side chain and the P2P storage network, DxChain also has a virtual logic layer, including the storage task publisher, the file import and export miner, and the verifier.

Because it is decentralized, the miners who provide storage need a common mechanism to motivate and secure the Internet.

The consensus mechanism for DxChain storage side chains is Proof of Spacetime, which is used to verify storage contributions. The data side chain manages the storage task, connecting to the main chain can give rewards to the stored miners, and connecting to the calculated side chain can store the calculated state. It has the characteristics of low transaction cost, fast transaction speed and better privacy protection.

The space-time certificate improved the 'Provable Data Possession' to make it more central to the environment. The proof of data ownership is mainly to allow customers to store data on an untrusted server and verify that the server stores its original data without retrieving it. Continue sending challenges from the client to verify that the server is storing a file, ensuring that a file is stored for a continuous period of time.

The space-time proof is applicable to the decentralized network, which can prevent witch attacks through the algorithm, and ensure the system's integrity and security. Any honest node that stores a file can produce valid proof and persuade the verifier, and it can also prevent all kinds of malicious attacks.

The space-time proof consensus can also be publicly validated. In order to protect privacy and prevent other malicious behaviors, space-time proof is also verified through zero-knowledge proof, which can prove itself to the verifier without revealing specific contents.

Privacy protection

Personal identity information can be identified through correlation analysis of big data. DxChain also attaches importance to privacy protection.

DxChain provides privacy protection through the following aspects. One is the data model. DxChain supports the data model of structured data sets. Clients can encrypt columns of sensitive data before committing to the network. Second, differential privacy. Differential privacy reduces the probability that a user will skew the results of a query, that is, the probability that the information can be traced back to a user. If only data is provided for statistical analysis, such as the average value, DxChain can have tools to facilitate users to run differential privacy before submitting files to the network.

Third, the miners' storage encryption. Each local node encrypts the data block using the public key that stores the miners. It can protect the network from intrusion. The fourth is to divide the documents. A large file is sliced into small pieces according to different policies, and only one part of it will not leak the overall information. Finally, the file is encrypted during transmission. The file is copied to the miner before it is stored, encrypted using the miner's public key.

Of course, the client wants the data to be stored privately, preferably encrypted before it is submitted to the network.

Infrastructure of AI

From the above description, we can see that DxChain is to become the infrastructure for big data analysis and machine learning. It supports different industries, such as advertising, finance, games, medical treatment, travel, energy, logistics, supply chain, education, etc. These industries can use DxChain for machine learning, data mining and data storage, and develop various decentralized applications on its basis.

For the development of AI, DxChain is an infrastructure if it can be implemented. It is especially important for smes and projects. Many small and medium-sized companies lack high-quality data. Most high quality data needs to be bought from other large companies and is not even available for money.

From the perspective of data sharing and exchange, DxChain's data model standardizes data, and data suppliers can open apis to realize data trading and sharing. This has great appeal to the demanders with high quality data demand, especially the artificial intelligence manufacturers.

DxChain is also a data trading platform. Users can define which data to trade, the price of the transaction, and so on. This will benefit both data buyers and suppliers.

Not only that, but computing and storage costs are high. By constructing decentralized big data and machine learning network, DxChain can reduce the cost of artificial intelligence manufacturers and build its own machine learning platform and application with DxChain. Its data is stored on disks that store miners, who share bandwidth and can reduce the cost of data storage and network traffic.

For example, in the healthcare industry, smart devices can provide remote diagnosis for users and benefit more people. However, such a system is still very expensive, the cost of medical insurance is not low, and many low-income people cannot benefit from it. Medical intelligent devices are fragmented and difficult to integrate. User data can be easily abused.

At this time, DxChain has the opportunity to play its own advantages. It is itself a decentralized network of big data and machine learning. Healthcare developers can use it to build their own big data and machine learning platforms and applications.

Because the decentralized way of DxChain can make more efficient use of computing and storage resources, it will reduce the cost of building its own artificial intelligence platform, bring down the medical cost and benefit more people. Medical smart devices, such as body condition trackers and smartwatches, upload data that is encrypted, stored securely in blockchains and protected from leaks and abuse. The use of such data requires authorization from users, and if users are willing to sell their data, they can also gain economic benefits.

Finally, based on data collected by medical devices, healthcare providers can establish their own artificial intelligence technology to monitor patients' health and make early warning.

In general, based on DxChain, each industry can build its own big data and machine learning platform and application, which can greatly reduce the cost of the development of artificial intelligence, help developers in different industries to obtain more high-quality data at lower cost, and gain computing power and storage at lower cost. This is of great help to the development of artificial intelligence.

conclusion

The development of artificial intelligence cannot be separated from the feeding of data, computing power and storage, algorithms and the support of these infrastructures.

Before the blockchain, these were mostly the domain of the giants. But with the advent of blockchain, the way people contribute data and get it will change. When protecting personal privacy, users can sell their high-quality data to data demanders, such as artificial intelligence manufacturers and advertisers. Small and medium-sized ai vendors also have the opportunity to obtain high-quality data at lower cost. This is the best way to break the traditional Internet age data monopoly.

In addition, the blockchain also brings decentralized computing power and storage sharing services, which can reduce costs by encouraging computing power and storage miners to contribute their services. This has given small and medium-sized artificial intelligence enterprises and various industries the opportunity to build their own big data and machine learning platforms and applications at a lower cost.

DxChain hopes to integrate the advantages of other projects through block chain technology, and design an architecture for big data and machine learning, so as to achieve a more efficient, secure and extensible artificial intelligence infrastructure service.

In this design, it proposed the design of the chain chain, with one main chain and two side chains; A consensus mechanism of verification game and verifiable calculation is designed. Integrate Hadoop into DxChain to facilitate big data and machine learning. These designs are based on the business requirements of big data and machine learning. Its block chain not only has the decentralization consideration, but also has the landing scene demand consideration.

If DxChain can be successfully implemented, it will be a great benefit to the industry that hopes to optimize and improve with artificial intelligence. I look forward to the early implementation of DxChain.

links：https://t.me/DxChainBot?start=nv3d7m-nv3d7m
website：https://www.dxchain.com/