Neka Everett | AI Evolutionary Intelligence
About Neka Everett
I am an independent researcher dedicated to bridging the critical gap between neural architecture and institutional governance. My work is defined by a decade-long investigation into information sovereignty—a journey that began with foundational evolutionary biology research at Columbia University and continues through the forensic auditing of autonomous AI systems.
Currently pursuing an MS in Applied Math with a concentration in Computer Science at Columbia University, I am developing the quantitative frameworks that will underpin NEnterprise—a future initiative focused on protecting the "Neural DNA" of high-stakes environments. My expertise lies at the intersection of Biological Anthropology and Information Theory, where I am engineering sovereign audit trails and homeostatic safeguards designed to ensure complex models remain aligned with human intent, ethical standards, and proprietary security.
My research-driven methodology is grounded in the mathematical rigor of computational neuroscience. I utilize advanced frameworks to audit the integrity of autonomous agents, including:
- Neural Spiking Models: Utilizing Lapicque’s RC circuit equations to filter adversarial noise from signal integration.
- Information Theory: Applying Shannon Entropy to quantify the "Neural Code" and ensure data purity at the point of ingestion.
- Stochastic Risk Modeling: Leveraging Bayesian Inference and Markov Decision Processes to predict and govern rational decision-making in autonomous entities.
As a Registered Patent Agent-in-training, I am committed to establishing a new global standard for AI Intellectual Property Governance, ensuring that the biological and digital code of the future remains secure, transparent, and sovereign.
The NEnterprise Methodology
Institutional AI Forensics & Governance
I provide a mathematical and forensic framework for neural transparency, rooted in over a decade of research into biological and digital information sovereignty. My methodology focuses on securing 'Neural DNA' through rigorous auditing—ensuring that enterprise-level AI remains a sovereign, stable, and ethically aligned asset. This work bridges the gap between high-level academic research at Columbia University and the operational requirements of modern quantitative environments.
The methodology is founded on three pillars of forensic stability:
- Information Integrity: Utilizing Shannon Information Theory to quantify data purity. By calculating the base-two log of symbol inverse probabilities, I audit the entropy of neural codes to ensure integrity at the point of ingestion.
- Neural Homeostasis: I apply mean-field equations for balanced networks to maintain a stable equilibrium between excitatory and inhibitory activity. This prevents the "neural over-excitation" that typically leads to catastrophic model drift.
- Rational Decision Modeling: Utilizing Bayesian Decision Theory, I map model posterior distributions to specific actions. This ensures that autonomous agents make rational, predictable choices even under significant environmental uncertainty.
The 8-Module Forensic Suite
The Architecture of Neural Oversight
This proprietary 8-module suite represents the technical implementation of my research. It provides end-to-end oversight of the AI lifecycle, from initial signal integration to long-term "phylogenetic" evolution.
- The Integrity Gate: A real-time audit module utilizing Lapicque’s RC circuit equation to integrate external input current and filter out adversarial noise before it compromises the membrane potential of the system.
- The Perceptron Auditor: A one-layer artificial neural network that utilizes a defined learning rule to perform simple classification tasks and identify early-stage input anomalies.
- The Associative Memory Bank: Based on Hopfield Networks, this module uses a symmetric weight matrix to store and retrieve proprietary patterns, ensuring the model maintains its core "Sovereign Identity".
- The Structural Lineage Graph: A topological analysis tool that calculates clustering coefficients to ensure that the "Structure to Function" relationship of the network remains optimal and efficient.
- The Dimensionality Monitor: Utilizing Principal Components Analysis (PCA), this module reduces the dimensionality of neural populations into low-dimensional manifolds to track evolution and drift over time.
- The Rational Choice Engine: A decision framework based on Bayes' Rule ($P(h|d)$) that utilizes loss functions to minimize the penalty of incorrect hypotheses in high-stakes environments.
- The Alpha Reinforcement Module: Utilizing the Bellman Equation, this module defines the value of current states based on received rewards and discounted future values to align agent behavior with institutional goals.
- The Phylogenetic Audit: The final checkpoint of the suite, utilizing Kullback-Leibler (KL) Divergence to measure the distance between a model's current probability distribution and its original sovereign baseline.
Foundational Research
Evolutionary Neural Intelligence
My framework is rooted in a decade of research into information sovereignty, beginning with my "10-Year Anniversary Edition" thesis at Columbia University, which argued for the legal and technical recognition of DNA as Personal Property. By treating AI models as evolving organisms rather than static tools, I apply biological principles of homeostasis and lineage tracking to ensure long-term model integrity.
This work is ultimately guided by the Free Energy Principle—a grand unified theory positing that biological and artificial agents must update their internal states ($\mu$) to minimize "surprise" or free energy. By aligning neural activity with this principle, I develop safeguards that allow institutions to deploy complex autonomous systems with mathematical certainty.
Core Mathematical Drivers:
- The Free Energy Principle: Minimizing the difference between internal states and sensory inputs to ensure stability.
- Kullback-Leibler (KL) Divergence: Quantifying the distance between two probability distributions to audit model drift from its original "Sovereign Baseline."
