Expand description
§RuVector Nervous System
Biologically-inspired nervous system components for RuVector including:
- Dendritic coincidence detection with NMDA-like nonlinearity
- Hyperdimensional computing (HDC) for neural-symbolic AI
- Cognitive routing for multi-agent systems
§Dendrite Module
Implements reduced compartment dendritic models that detect temporal coincidence of synaptic inputs within 10-50ms windows. Based on Dendrify framework and DenRAM RRAM circuits.
§Example
use ruvector_nervous_system::dendrite::{Dendrite, DendriticTree};
// Create a dendrite with NMDA threshold of 5 synapses
let mut dendrite = Dendrite::new(5, 20.0);
// Simulate coincident synaptic inputs
for i in 0..6 {
dendrite.receive_spike(i, 100);
}
// Update dendrite - should trigger plateau potential
let plateau_triggered = dendrite.update(100, 1.0);
assert!(plateau_triggered);§HDC Module
High-performance hyperdimensional computing implementation with SIMD-optimized operations for neural-symbolic AI.
§Example
use ruvector_nervous_system::hdc::{Hypervector, HdcMemory};
// Create random hypervectors
let v1 = Hypervector::random();
let v2 = Hypervector::random();
// Bind vectors with XOR
let bound = v1.bind(&v2);
// Compute similarity (0.0 to 1.0)
let sim = v1.similarity(&v2);§EventBus Module
Lock-free event queue system for DVS (Dynamic Vision Sensor) event streams with 10,000+ events/millisecond throughput.
§Example
use ruvector_nervous_system::eventbus::{DVSEvent, EventRingBuffer, ShardedEventBus};
// Create event
let event = DVSEvent::new(1000, 42, 123, true);
// Lock-free ring buffer
let buffer = EventRingBuffer::new(1024);
buffer.push(event).unwrap();
// Sharded bus for parallel processing
let bus = ShardedEventBus::new_spatial(4, 256);
bus.push(event).unwrap();Re-exports§
pub use compete::KWTALayer;pub use compete::LateralInhibition;pub use compete::WTALayer;pub use dendrite::Compartment;pub use dendrite::Dendrite;pub use dendrite::DendriticTree;pub use dendrite::PlateauPotential;pub use eventbus::BackpressureController;pub use eventbus::BackpressureState;pub use eventbus::DVSEvent;pub use eventbus::Event;pub use eventbus::EventRingBuffer;pub use eventbus::EventSurface;pub use eventbus::ShardedEventBus;pub use hdc::HdcError;pub use hdc::HdcMemory;pub use hdc::Hypervector;pub use hopfield::ModernHopfield;pub use plasticity::eprop::EpropLIF;pub use plasticity::eprop::EpropNetwork;pub use plasticity::eprop::EpropSynapse;pub use plasticity::eprop::LearningSignal;pub use routing::BudgetGuardrail;pub use routing::CircadianController;pub use routing::CircadianPhase;pub use routing::CircadianScheduler;pub use routing::CoherenceGatedSystem;pub use routing::GlobalWorkspace;pub use routing::HysteresisTracker;pub use routing::NervousSystemMetrics;pub use routing::NervousSystemScorecard;pub use routing::OscillatoryRouter;pub use routing::PhaseModulation;pub use routing::PredictiveLayer;pub use routing::Representation;pub use routing::ScorecardTargets;pub use separate::DentateGyrus;pub use separate::SparseBitVector;pub use separate::SparseProjection;
Modules§
- compete
- Winner-Take-All Competition Module
- dendrite
- Dendritic coincidence detection and integration
- eventbus
- Event Bus Module - DVS Event Stream Processing
- hdc
- Hyperdimensional Computing (HDC) module
- hopfield
- Modern Hopfield Networks
- integration
- Integration layer connecting nervous system components to RuVector
- plasticity
- Synaptic plasticity mechanisms
- routing
- Neural routing mechanisms for the nervous system
- separate
- Pattern separation module implementing hippocampal dentate gyrus-inspired encoding