Spiking Neurons
Process time the way brains do.
Spiking Neural Networks for ultra-low-latency temporal pattern processing. LIF, Izhikevich, and adaptive neurons. STDP learning and BPTT training. Ideal for tick streams, sensor data, and log anomaly detection.
Capabilities
What it does.
Every capability is production-grade. No flags, no betas — these ship the day you adopt Minds.
LIF · Izhikevich · adaptive
Three neuron families, configurable per-ensemble.
STDP learning
Spike-Timing-Dependent Plasticity for unsupervised pattern discovery.
BPTT training
Backprop-Through-Time for supervised tasks.
Compile-once execution
Define network declaratively, compile to runtime once, step many.
State introspection
Inspect voltages, spikes, and synaptic weights at any timestep.
Cognitive integration
SNN ensembles feed the Cognitive Cycle's temporal pre-processor.
API
A few lines is all it takes.
The SDK reflects the conceptual model directly. No glue code, no orchestrator, no learning curve beyond the data shape.
exampletypescript
// Build a spiking network.
const net = minds.snn.compile({
ensembles: [
{ name: "input", n: 100, model: "LIF" },
{ name: "hidden", n: 200, model: "Izhikevich" },
{ name: "output", n: 10, model: "adaptive" },
],
connections: [
{ from: "input", to: "hidden", learn: "STDP" },
{ from: "hidden", to: "output", learn: "STDP" },
],
});
// Step the simulation.
for (let t = 0; t < 1000; t++) {
net.input(sensor.next());
net.step();
if (t % 100 === 0) console.log(net.spikes("output"));
}Specs
What it costs, what it guarantees.
Performance
- Spike throughput
- 1.4M spikes/sec per node
- Step latency
- <1ms
- Learning rules
- STDP · BPTT
- Neuron models
- LIF · Izhikevich · adaptive
Guarantees
- Durability
- fsync per WAL commit
- Isolation
- MVCC · snapshot
- Audit
- BLAKE3 Merkle chain
- Encryption
- AES-256-GCM per-namespace
- Concurrency
- 100K+ ops/sec
Build it on Minds.
Start with the SDK. Ship in an afternoon. Run anywhere — Akasha Cloud, on-prem, or air-gapped.