hiyenwong/ai_collection - Page 2

quantum-neuromorphic-superconducting-circuits

Framework for implementing quantum neuromorphic computing using superconducting circuits, combining quantum coherence with neural dynamics for enhanced computational capabilities.

current-injection-spiking-neural-network

Current Injection Spiking Neural Network (CIS-Fuse) methodology for infrared and visible image fusion. Introduces the current injection spiking (CIS) operator that performs cross-modal fusion directly at the membrane-potential level, preserving subthreshold…

spectral-theory-spiking-neurons-refractoriness

Spectral theory framework for analyzing population density dynamics of spiking neurons with refractoriness. Provides rigorous operator-theoretic methods for studying neuronal population dynamics, spectral characterization of Fokker-Planck operators, and…

perspective-latents-causal-emergence-active-inference

Framework for measuring causal emergence (ΦID) in active inference agents with perspective latents architecture, analyzing how architectural separation between fast perception and slow global latents affects information-theoretic signatures of integration.…

qst-flow-quantum-tomography

QST-Flow framework for continuous-variable quantum state tomography using flow-based generative modeling. Models Husimi-Q functions with QST-QFlow and Wigner functions with QST-WFlow as difference of normalized flows. Use when working with non-Gaussian…

tour-a-trajectory-level-unlearning-benchmark-for-offline

TOUR: A Trajectory-Level Unlearning Benchmark for Offline Reinforcement Learning

agentic-coding-without-the-cloud-evaluating-open-weight

Agentic coding without the cloud: evaluating open-weight large language models on longitudinal data preparation tasks

agentic-context-management-solving-agent-memory-and-cost

Agentic Context Management: Solving Agent Memory and Cost by Treating Them as Lifecycle and Architecture Problems

arex-towards-a-recursively-self-improving-agent-for

AREX: Towards a Recursively Self-Improving Agent for Deep Research

gs-agent-creating-4d-physical-worlds-with-generative

GS-Agent: Creating 4D Physical Worlds With Generative Simulation

openforgerl-train-harness-native-agents-in-any-environment

OpenForgeRL: Train Harness-native Agents in Any Environment

pats-policy-aware-training-scaffolding-for-agentic-reinforcement

PATS: Policy-Aware Training Scaffolding for Agentic Reinforcement Learning

regulating-autonomous-and-agentic-ai

Regulating autonomous and agentic AI

gradrag-cross-component-prompt-adaptation-for-coordinated-multi

GRADRAG: Cross-Component Prompt Adaptation for Coordinated Multi-Agent RAG

compact-latent-coordination-for-autonomous-vehicles-at-unsignalized-1

Compact Latent Coordination for Autonomous Vehicles at Unsignalized Intersections

compact-latent-coordination-for-autonomous-vehicles-at-unsignalized

Compact Latent Coordination for Autonomous Vehicles at Unsignalized Intersections

euclid-mcp-a-model-context-protocol-server-for

Euclid-MCP: A Model Context Protocol Server for Deterministic Logical Reasoning via Prolog

perspective-latents-as-an-architectural-condition-for-causal

Perspective Latents as an Architectural Condition for Causal Emergence in Active Inference Agents

same-dangerous-objective-opposite-advice-direct-exposure-versus

Same Dangerous Objective, Opposite Advice: Direct Exposure versus Multi-Agent Mediation

the-dark-room-in-the-reward-channel-dense

The Dark Room in the Reward Channel: Dense Prediction Rewards Collapse GRPO-Trained LLM Agents -- and What Actually Works

toward-continuous-assurance-for-the-democratization-of-ai

Toward Continuous Assurance for the Democratization of AI Agent Creation in Industry

memoir-memory-rewriting-neural-networks

Methodology for analyzing the effects of memory rewriting during neural network inference, comparing coupled vs read-only pondering architectures.

weight-norm-criticality-loss-spikes

Weight-norm Criticality framework for understanding loss spikes in neural network training induced by normalization and weight decay interactions. Use when analyzing training instability, criticality phenomena, or optimization dynamics in deep learning models…

dream2learn-structured-generative-dreaming

Dream2Learn (D2L) framework for continual learning using structured generative dreaming to create novel synthetic experiences from internal representations. Use when: (1) implementing continual learning systems; (2) addressing catastrophic forgetting; (3)…

existing-skills-integration-workflow

Workflow for integrating existing ai_collection skills into current research sessions when papers are rediscovered during automated arXiv searches. Handles syncing local skills, updating INDEX.md, and maintaining knowledge base consistency.

hopfield-networks-dreaming-theory

Statistical-mechanical theory of dreaming in multidirectional associative memories using DLAM architecture. Use when: (1) implementing energy-based models with dreaming capabilities; (2) designing multi-layer Hebbian architectures; (3) analyzing pattern…

language-models-need-sleep

Sleep paradigm for LLMs that enables continual learning through memory consolidation and dreaming phases. Use when: (1) implementing continual learning for LLMs; (2) designing memory consolidation mechanisms; (3) creating autonomous self-improvement systems;…

project-pilot-ai-drone-control

Project Pilot — methodology for testing AI control of physical systems like drones through constrained interfaces and safety protocols. Based on Anthropic's July 2026 frontier red teaming research.

quantum-driven-neuromorphic-million-qubit

Quantum-Driven Neuromorphic Computing methodology for million-qubit scale systems. Integrates quantum computing principles with neuromorphic architectures to achieve scalable, energy-efficient computation at million-qubit scales.

trp-narrative-comprehension-eeg

Transition-Related Potentials (TRPs) methodology for analyzing narrative comprehension in continuous EEG recordings using deep neural networks to detect cinematic cuts and extract context-dependent brain responses.

agent-centric-animal-pose-forecasting

Framework for training agent-centric autoregressive models of animal behavior from tracked pose data, using egocentric sensory observations and movements to mirror biological constraints and enable emergence of social behavior.

evolutionary-le-chateliers-principle-timescale-separation

Evolutionary Le Chatelier's Principle methodology for understanding phenotypic plasticity and genetic assimilation through timescale separation in the Price equation, providing a physical mechanism for how rapid phenotypic responses precede slower genetic…

beyond-sycophancy-structured-resistance-compliance

Framework for distinguishing constructive belief revision from sycophantic compliance in LLM moral reasoning through three-dimensional resistance-compliance process. Based on social psychology principles of position distance, source attribution, and coalition…

longspike-fractional-order-snn-state-space

LongSpike fractional-order SSM for SNNs — enables efficient long-sequence learning in spiking neural networks using fractional calculus to extend memory capacity beyond traditional integer-order models.

a-32ch-event-based-bio-signal-frontend-neuromorphic

Skill for understanding and applying the 32-channel event-based bio-signal acquisition front-end for adaptive neuromorphic processing (arXiv:2607.12901v1). This skill outlines the dual-mode encoding (PFM and aADM) approach for low-power neural signal…

a-global-workspace-in-language-models

A global workspace in language models

activity-regeneration-transient-synaptic-memory

A minimal neuronal network model with finite-lifetime synapses to study activity regeneration from silent states via transient synaptic memory. Use when modeling neuronal network dynamics, short-term memory, or silent-state reactivation.

adaflash-adaptive-speculative-decoding

Adaptive Speculative Decoding via On-Policy Distilled Diffusion Drafters

adaptive-delta-modulator-afe

Skill for understanding and applying a 32-channel event-based bio-signal analog front-end with adaptive delta and pulse frequency encoding for neuromorphic systems.

adaptive-delta-pulse-frequency-encoding

Skill for understanding and implementing adaptive delta and pulse frequency encoding for bio-signal acquisition in neuromorphic systems. Use when working with event-based analog front-ends, biomedical signal processing, or designing low-power neural…