quantum-optical-neuron

name	quantum-optical-neuron
description	Quantum optical neuron methodology — camera-free image classification via Hong-Ou-Mandel interference of spatially programmable single photons. Two-photon coincidences directly report overlap between input image mode and learned template. Use when building neuromorphic quantum photonic processors, photon-starved imaging systems, or quantum-classical hybrid inference pipelines.
metadata	{"arxiv_id":"2603.28879","published":"2026-03-30","authors":"Unknown","tags":["quantum","neuromorphic","photonics","imaging","neuroscience"]}

Quantum Optical Neuron

Core Concept

Camera-free quantum-optical image classifier using Hong-Ou-Mandel (HOM) interference. Input images are encoded as spatial modes of single photons; two-photon coincidences directly measure the overlap between input and a learned template, replacing pixel-resolved acquisition with a single global measurement.

Mathematical Framework

HOM Interference as Inner Product:

• Input state: |ψ_in⟩ = spatial mode encoding image
• Template state: |ψ_template⟩ = learned reference mode
• Two-photon coincidence rate: P_coincidence ∝ |⟨ψ_in|ψ_template⟩|²
• Direct measurement of image-template similarity without pixel-by-pixel scanning

Single-Perceptron Quantum Optical Neuron:

• Single tunable beam splitter with spatial light modulator
• Threshold on coincidence rate → binary classification

Shallow Quantum Neural Network:

• Two-neuron architecture with programmable weights
• Cascaded HOM interferometers for multi-class classification

Key Properties

1. Resolution Independence: Performance insensitive to input resolution under fixed measurement budget — fundamentally different from classical pixel-scaling
2. Photon Efficiency: Operates in photon-starved regimes where classical cameras fail
3. Noise Robustness: Strong robustness to experimental noise from quantum interference properties
4. Hardware Simplicity: Minimal optical components vs. full quantum optical processors

Usage Patterns

Pattern 1: Photon-Starved Imaging Classification

When classical imaging SNR is insufficient (remote sensing, biological microscopy):

1. Encode image as spatial photon mode via SLM
2. Prepare template states for each class
3. Measure HOM coincidence rates against each template
4. Classify by maximum coincidence rate

Pattern 2: Neuromorphic Quantum Photonic Processor

For building scalable quantum-neuromorphic systems:

1. Implement single-perceptron quantum optical neurons
2. Connect via tunable beam splitters (synaptic weights)
3. Cascade for shallow network inference
4. Train template states via classical optimization loop

Pattern 3: Energy-Efficient Inference

For low-power edge inference with photon-level data:

1. Use HOM interference as physical-layer inner product computation
2. Avoid ADC + digital processing pipeline entirely
3. Achieve classification at measurement layer directly

Implementation Notes

• Platform: Linear optical quantum computing with SPDC sources or quantum dots
• Encoding: Spatial light modulator for mode preparation
• Detection: Single-photon avalanche diodes (SPADs) for coincidence counting
• Measurement Budget: Fixed number of coincidence measurements; optimize allocation across classes

Cross-References

• [[quantum-memristor-vacuum-one-photon]] (2503.02466) — Quantum memristors for memory-dependent quantum neurons
• [[quantum-snn-fusion]] — Quantum-SNN hybrid architectures
• [[neuromorphic-quantum-computing]] — Broader neuromorphic quantum computing patterns

Activation Keywords

• quantum optical neuron
• HOM interference classification
• camera-free quantum imaging
• 量子光学神经元
• photon-starved image classification
• neuromorphic quantum photonic
• quantum perceptron optical
• Hong-Ou-Mandel neural network