Green Blockchain Cryptocurrency
Much has been said about blockchain cryptocurrency use of unsustainable electric power for its Proof of Work (PoW) transaction validating computations. Proof of Stake (PoS) calculations, which attempt to reduce the validating computational burden of PoW approaches have proved difficult to implement.
QAI.ai believes in the future of blockchain cryptocurrency and is focused on making it a green technology, which does not use excessive amounts of electrical power and also prevents centralization of transaction verification. We have a three stage plan to implement cryptocurrency greening.
Machine Learning Background
All current machine learning algorithms, whether they be supervised, unsupervised or reinforcement learning, are between 25 and 40 years old. In a discussion I had with my former GTE colleague Dr Rich Sutton, we stated that nothing has changed in current machine learning algorithms, only computer processing power, memory and storage costs have changed. Rich is the co-founder of reinforcement learning algorithms along with Dr Andy Barto at University of Massachusetts at Amherst. Rich was brought in by Google's Geoffrey Hinton, (co-founder of supervised learning algorithms with the late David Rummelhart) to help Google develop reinforcement learning technology to address fundamental bias limitations of supervised learning. Supervised learning systems are inherently biased independent of data or training because in order to get them to generalize properly one must overfit the training inputs, which leads to bias in these systems.
QAI Digital Physics Machine Learning - A New Machine Learning Paradigm
QAI is a fundamentally new form of machine learning based upon Digital Physics. Digital Physics holds that the probabilistic nature of quantum mechanics and general relativity are compatible with all notions of computability. The hypothesis was pioneered in Konrad Zuse's book Rechnender Raum (translated by MIT into English as "Calculating Space", 1970) holds that particles, such as an electron, may be viewed as switching from one quantum state to another, i.e. like a bit change from one value (0) to another (1). In this way every fundamental quantum state and their general relativity project into four dimensional space are simply information and every quantum or relativistic change is a change in information.
QAI first implemented novel Digital Physics algorithms in 1997. The accuracy of QAI Digital Physics was refined and calibrated by applying it to difficult computational challenges in quantum chemistry. The idea being that if the QAI Digital Physics implementation is correct it should be possible to test it by experimentally know quantum chemistry calculation, including solving for the structure of the Hydrogen Molecular Ion, without use of the Born Oppenheimer approximation upon which all quantum chemistry is based. The dynamic ab initio solution to the Hydrogen Molecular Ion without the Born Oppenheimer approximation is shown below. This solution was first demonstrated during QAI validations in 2003.
Many validations of QAI Digital Physics calculations followed on much larger molecular systems at femtometer resolutions. Due to the high resolution QAI enabled understanding of supramolecular interactions, which were formerly described by chemistry ad hoc rules and explanations. An overview of the validation and refinement period of QAI Digital Physics, lasting over a decade, is discussed in this video (to come).
Following the QAI validation and refinement period, QAI fundamental Digital Physics algorithms, which by definition should apply generally to any computational problem, were extracted and implemented for general purpose computation. Specifically, a new form of machine learning, which is neither supervised, unsupervised or reinforcement learning algorithm. In this new QAI Digital Physics machine learning, machine learning is the dynamic formation and change to information, which is represented within QAI as Digital Physics information molecular structures and their complexes. Using QAI dynamic convergence, which is O(n) time on a serial computer and O(1) on a parallel machine, QAI solves for complex machine learning problems by rapidly transducing them into their equivalent dynamic information molecular signatures. These information molecular signatures may be combined using QAI Digital Physics to solve for complex optimization problems without the need to resort to partial differential equation methods. QAI Digital Physics represents all information and knowledge in the form of vibrating quantum strings whose projections are viewed in four spatial dimensions within a continuous computational space shown below (to come).
QAI Digital Physics Computation Summary
QAI implements a new Digital Physics form of general purpose computation with implementations in both software and superconducting exotic computer hardware. When applied to implementing machine learning calculations it generates dynamic information molecular signatures, whose structural analogues provide for unbiased generalization and nearly instantaneous fully autonomous machine learning without training. QAI may revolutionize computation by successfully adapting the current physics standard model to general purpose computation. The standard model of the fundamental interactions, less the Higgs Boson, is below.
QAI is currently implemented as a Peer to Peer (p2p). The QAI p2p architecture was first deployed within a Beowulf supercomputing cluster in 1995 and pre-dates all current p2p architectures. The QAI system infrastructure is based upon concepts derived during the design of the National Security Agency (NSA) Embroidery world wide secure network backbone and the fault immune messaging system of both the LGM-118 Peacekeeper thermonuclear weapon system and the coding used by the U.S. military employed Extremely Low Frequency (ELF) transmissions used by US Ohio class trident equipped submarines. Consequently, QAI is inherently secure and enabled with multi-level high assurance compartmentalization. We believe QAI applied to Blockchain Cryptocurrency addresses the current limitations of these systems.
Cryptocurrency Stage 1
Stage one involves making the PoW calculation more efficient. We are implementing stage one using Quantum Artificial Intelligence (QAI) "Learn to Hash" (L2H) technology. Current PoW hash code generation use data independent methods, whose mathematics are straightforward and easily implemented but are computationally intensive. This results in an unsustainable drain on the world wide power grid.
QAI L2H methods implement unbiased, real-time machine learning, which requires no training. QAI L2H uses high order differential cryptanalysis to reduce the complexity of the current PoW calculation by a factor of of up to two orders of magnitude, i.e. 100x. QAI does this by "windowing" the PoW mining calculation, which reduces its overall computational complexity. Although this does not fundamentally address the PoW calculation long term unsustainability and other issues regarding overall blockchain challenges, we believe it is a good start to demonstrate the power of QAI L2H machine learning. Since QAI requires access to the blockchain and collective control of the ASIC miners, a secondary goal is prediction of cryptocurrency volatility and the identification of cryptocurrency attempts at manipulation and possible prevention. We expect deployment of QAI L2H data dependent mining by the end of 2018. We currently use a test bed of 11 Bitminer S9 ASIC miners all controlled by QAI. We anticipate deployment in a 3 megawatt ASIC facility, which is under construction in the tundra of the Montana wilderness using inexpensive hydroelectric power.
Cryptocurrency Stage 2
QAI L2H methods in stage 1 function as conventional PoW miners, which appear as extremely fast computational ASIC mining machines due to the greatly reduced PoW QAI windowed mining calculations. In stage 2 we incorporate the QAI p2p system as part of the blockchain itself. This incorporation replaces current PoW and Proof of Stake (PoS) alternatives to mining with a new QAI Digital Physics decentralized blockchain transaction validation calculation, which executes in O(n) time and O(1) time using a plurality of QAI augmented blockchain nodes. The unique change here is that the QAI replaces the current unsustainable PoW and intractable PoS calculations with a sustainable QAI Digital Physics information molecule calculations, which is now systemic to each blockchain p2p node. This is a soft fork exists alongside existing mining calculations but now within the blockchain nodes themselves implement the basis for true blockchain self-managed p2p node autonomy, ultimately without need for conflicting blockchain miner requirements. The stage 2 QAI Blockchain soft fork builds on the stage 1 QAI L2H system but with a different blockchain validating calculation that is neither PoW or PoS and that is the basis for Blockchain Green sustainability. The name of the stage 2 project is internally called "Gunslinger" to differentiate it from stage 1 requirements.
Cryptocurrency Stage 3
QAI Blockchain stage 3 is a hard fork that implements full blockchain node automonomy and elimination of unsustainable mining calculations. In stage 3 the QAI green sustainable information molecule blockchain transaction validation calculation is decentralized across all the blockchain nodes. This decentralization is immune to failure due to the fine grained distributed nature of the QAI Digital Physics calculation. If a node or many nodes go offline QAI decentralized Digital Physics calculations reconstitute the information from the offline nodes. This makes the p2p blockchain QAI system in stage 3 survivable against attack since QAI in its Blockchain p2p implementation is also now managing the decentralized blockchain node structure with the specific goal of survivability and immunity from attack or outside interference. In this stage QAI uses its agile channel hoping self-encrypted communications over IP to provide for full blockchain network autonomy and survivability. The goal of stage 3 is full cryptocurrency international autonomy.