KorrTecx Project
Neural Research Lab
Neural Research Lab
The future of aligned machine intelligence.
Next-Gen AI - A reverse-engineered, bio-plausible neural model that emulates 'human-type' modalities, leveraging holistic memory schemas, continual learning, imagination & intelligence.
Research Initiative - Reimagined from first principles, created to further our understanding of the human condition, neurodegeneration and maybe even consciousness.
Humans are bio-electrochemical machines, there's no magic... only complexity.
A Functioning Bio-Intelligence
Neural processing takes time... yet we do not lag behind reality.
Real-time perception/ recognition of an experience requires predictive modelling constructed from existing priors. New learning is then engrained relative to those priors, incrementally enhancing the world model through experience.
Three Layered Approach...
#1 Bio Simulation Engine
Bespoke 4D CAD & simulation engine, a mesh of 3D guides defines the spatial characteristics of the neural structures and laminate layers. An embedded octree mesh enables volumetric bio-electrochemical functionality.
Algorithms simulate transmitter & modulator gate uptake, dissipation, accumulation, neuro/ synaptogenesis, infinite neuron types & arbour descriptors, maturation, growth, sleep & many other bioprocesses.
#2 Biomimetic Model
Taking the human connectome & nervous system as a template the model comprises the major structures including neocortex lobes, thalamus, hippocampus, limbic & brain stem structures.
Connectomics are defined by evolutionary algorithms that evolve & adapt the model over time. Long-range inter-structure (white matter) axons, the medium/ short-range (grey matter) tracts within the semantic laminate layers & localised structures like dendritic arbours and hubs are all simulated.
#3 Neural Code
The complex electrochemical waves of activity that flow through the model comprise the neural code that runs on the virtual architecture of the spatiotemporal processor (connectome).
The self-sustaining thalamocortical rhythm encodes a holistic memory schema & drives temporal pattern completion through inertia & predictive encoding, enabling secondary mechanisms to manifest attention & alignment.
Generating over a billion equivalent 'parameters' per second.
Understanding is achieved by recognising commonalities between similar concepts.
~ & ~
New discoveries are realised by recognising commonalities between diverse concepts.
Bio Plausibility - Methodology
Leveraging empirical research & tens of thousands of simulation runs to refine the functionality whilst keeping as true as possible to the biological template. Simulating neurodegenerative conditions within the context of the model, dementia, schizophrenia, MND, etc, helps ground the design... if the architecture & mechanisms are accurate, it should be liable to similar phenomena.
Quick Deep Dive
As an introduction to the KorrTecx biomimetic neural model, a brief description of the initialisation sequence required to build a full connectome model.
Each model is unique, once initiated it needs no further user input, it grows, scales & matures according to a set of algorithms, becoming more salient & intelligent through experience over time.
Neural Theory - Abstract
The beauty is that the emergent connectome defines both the structural hardware and the software. The brain is more akin to a clockwork watch or a Babbage engine than a modern computer. The design of a cog defines its functionality. Data is not passed around within a watch, there is no software; but complex calculations are still achieved. Each module does a specific job, and only when working as a whole can the full and correct function be realised.
