QAI SPECTRE™

Patent Pending Super Turing Machines

Analog Self-Organizing Recurrent Neural Networks

Current and future digital machines including any possible scalable quantum computers are P-Space reducible, which means they are constrained to the Turing limit.
Conversely, QAI SPECTRE™ analog recurrent neural network super Turing machines are not. QAI SPECTRE™ machines compute precisely with real/irrational numbers, which Turing limited machines only approximate.
— Lars Wood - CEO QAI and inventor QAI SPECTRE™ super Turing Machines

Digital machines described by the Turing Limit are restricted to computations involving rational numbers. Digital machines are further defined by the Church-Turing Thesis as having processing separate for memory. Super Turing machines calculate using real numbers (includes direct implementation of imaginary complex numbers, for example). This results in vast more degrees of problem solving freedom. QAI SPECTRE™ super Turing machines are implemented as dynamically reconfigurable analog recurrent neural networks where memory and processing are unified.

The patent pending QAI SPECTRE™ super Turing architecture is implemented as a self-healing GaAs/GaN on Diamond 4D multi-chip module. Through modification of gate level structures, as described in the QAI SPECTRE™ patent, the analog QAI SPECTRE™ neurons implement dynamic wireless sub millimeter wave frequency interconnect. This not only results in QAI SPECTRE™ neuron high fan in and fan out, it also results in optimal high performance problem solving using self-organizing neural computation.

QAI SPECTRE™ asynchronous neurons function at 3 THz frequencies to implement blindingly fast super Turing neural computing that far exceeds digital computation through Real number computation problem solving degrees of freedom. QAI SPECTRE™ enables adaptive AI and machine learning solutions, which are currently infeasible with any digital machine because QAI SPECTRE™ exceeds the theoretical Turing digital computation limit. A typical QAI SPECTRE™ module draws less than one Joule of energy and its recurrent neural networks can exceed 500 quadrillion operations per second.

SPECTRE enables adaptive AI solutions for applications in Robotics, UAVs and intelligent self-driving cars. However QAI SPECTRE™ is much more. It is a general new computational solution that exceeds digital computation for applications in cryptanalysis, big data analysis and other traditional digital computation applications.

QAI believes that QAI SPECTRE™ technology fundamentally obsoletes and is a full replacement for digital technology and will ultimately change the world computation.