Diffractive Deep Neural Networks (D2NN)

Cross-cuts: Communications
last updated Thu May 07 2026 00:00:00 GMT+0000 (Coordinated Universal Time)
Spatial Light Modulators (SLMs)Holographic / Volumetric DisplaysFree-Space PhotonicsMetasurfaces & Flat OpticsDiffractive…

Architecture class within ONNs.

Physics / mechanism

Origin: Lin et al. (Ozcan group, UCLA) demonstrated the first all-optical D2NN classifier in Science 2018, with cascaded 3D-printed phase plates performing MNIST-class classification at the speed of light through the stack. The architecture has since spread to USTC, Stanford, CMU, and many others.

Key parameters:

Distinguished from sibling architectures

ArchitectureCompute primitiveWhere it livesKey trade-off
D2NN (this page)Diffraction through phase masksFree spaceMassively parallel, but precision compounds; reconfigurability tied to SLM speed
MZI mesh ONN (Lightmatter, Lightelligence)Guided-wave Mach-Zehnder phase shiftersSilicon photonic ICFoundry-fabable; but O(N²) phase shifters limits scale
Photonic tensor coresWavelength-multiplexed multiply-accumulateSilicon photonic ICHigh throughput; precision still 4-6 bit
Photonic reservoirFixed nonlinear dynamical systemFree-space or fibreGood for time-series / ODEs; limited generality, no commercial breakout
Analog optical (Salience)PCM crossbar with optical readoutHybrid PCM + photonicFoundry-adjacent; precision-bound

Why this matters now (2026)

Companies using

Connected ideas

Sources

Frontier (open questions)

See frontmatter frontier: block.

Reading list

Related concepts

Frontier questions