The invention of Bitcoin in 2008 symbolized the advent of cryptocurrency. The succeeding years witnessed a dramatic growing of the blockchain technology. The introduction of Ethereum opened the path for supporting complex business behaviors with cryptocurrencies. The overall momentum of development, however, is now facing a series of key challenges. These include: (1) slow transaction speed; (2) programming barrier of smart contracts; (3) lack of security in smart contracts; and (4) inflexibilities in managing and updating blockchains. Designed to be the new generation blockchain, MATRIX leverages the latest artificial intelligence technology to resolve the abovementioned challenges. The fusion of blockchain and AI technologies enables MATRIX to build a revolutionary cryptocurrency, which support significantly boosted transaction speed, superior accessibility to general users, enhanced security under malicious attacks, and highly flexible operations.
The challenges facing today’s blockchains have to be resolved before the ideal of cryptocurrency can really become reality. We believe the artificial intelligence technology, which has received an unprecedented growth in the past decade, provides out-of-box solutions to address the challenges. MATRIX is designed to be an intelligent chain to unleash the potential power of the blockchain technology. In this section, we briefly review the objectives of MATRIX.
Although smart contracts give blockchains the essential capability to handle scaled commercial behaviors, they need the users to be able to write programs in a given programming language. With MATRIX, no programing expertise is needed any more for designing smart contracts. The unique code generation technique of MATRIX allows automatic conversion of an abstract description of a smart contract into an executable program. Matrix only requires users to input the core elements (e.g., input, output, and transaction conditions) of a contract with a scripting language. Then a code generator based on a deep neural network is able to automatically convert the script into an equivalent program.
Smart contract programs may call functions offered by the host system and/or third-party libraries. Also, programs running on different computers in a distributed framework do not provide any guarantee for execution time. Such openness and decentralization are the reflection of the essential spirit of blockchains, but give birth to various sources of security threats. In fact, the lack of security is plaguing the smart contracts. The MATRIX blockchain is equipped with a power AI security engine consisting of four major components, 1) a rulebased semantic and syntactic analysis engine for smart contracts, 2) a formal verification toolkit to prove the security properties of smart contracts, 3) an AI-based detection engine for transaction model identification and security checking, and 4) a deep learning based platform for dynamic security verification and enhancement
Today all public chains are suffering from the problem of long transaction latency and low transaction throughput. Specifically, it takes over 30 minutes for Bitcoin to finish one transaction, while the transaction throughput of Ethereum is only 10 Transfer Per Second (TPS). In fact, a blockchain depends on a P2P network to validate transactions. Since a transaction needs to be broadcasted to all nodes in a network, the overall latency has to increase as long as more nodes are joining the network. MATRIX resolves the problem by dynamically selecting a delegation network in which all nodes are voted as delegates of others. All Proof-of-Work (PoW) processing is only allocated inside the delegation network, which only incurs a much smaller latency due to the smaller number of nodes. The selection process is random in the sense that a node is selected with a probability proportional to its Proof-of-Stake (PoS). The online version of MATRIX will support a throughput of 100,000 TPS.
MATRIX is designed to be a highly flexible blockchain. The flexibility is twofold. First, MATRIX offers access control and routing services so as to allow seamless integration private chains into a common public chain. Such a feature meets the requirements many industry and government players for authorization, while at the same time allows necessary information flow from a public chain to a private one and vice versa. Second, MATRIX uses a reinforcement learning framework to optimize its parameters (e.g., consensus mechanism, and transaction configuration) in an evolutionary manner. The optimization paradigm ensures dynamic updating of parameters for near-optimal performance without the risk of incurring hardfork.
Perhaps the most criticized part of cryptocurrency is the “waste” of energy in the mining computations. Although it is essential to attach physical value to the cryptocurrency, the mining process does not make any sense out of the world of digital currency. The problem is even worse when now over 70% of the total computing power around of world is dedicated to mining Bitcoins and others. MATRIX introduces a new mining mechanism in which miners perform the Markov Chain Monte Carlo (MCMC) computation, which is an essential tool for Bayesian reasoning.
MCMC based Bayesian computing plays a fundamental role in numerous big data applications such as gene regulatory network, clinical diagnosis, video analytics, and structural modeling. As a result, a distributed network of MCMC computing nodes provide the power for solving real-world compute-intensive problems and thus build a bridge between the values in the physical and virtual worlds.
- Steve Deng - Chief AI Scientist
- Bill Li - Chief Network Architect
- Tim Shi - Chief Chip Scientist
- Owen Tao - CEO
- John Zhu - Senior VP
- Ethan Tian - Chief R & D Engineer
- Dr. Donglin Wang - Advisers
- Tony Surtees - Advisers
- Ouyang Hongwu - Advisers
- Dr. Jinyang Wang - Advisers