Optical neural networks: The 3D connection

Dinç, Niyazi Ulaş; Brunner, Daniel

doi:10.1051/photon/202010434

Cited by 27 publications

(17 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Volume holograms have been of great interest for parallel optical interconnects and data storage [27] as a 3D GRIN connection scheme [28]. When a hologram is optically recorded in a photosensitive medium by interfering a reference (𝐸 ) and object (𝐸 ) beam, the recorded hologram is given by:…”

Section: Experimental Demonstrations a Volume Hologramsmentioning

confidence: 99%

Direct (3+1)D laser writing of graded-index optical elements

Porté

Dinç

Moughames

et al. 2021

Optica

Self Cite

View full text Add to dashboard Cite

We propose the single-step fabrication of (3+1)D graded-index (GRIN) optical elements by introducing the light exposure as the additional dimension. Following this method, we demonstrate two different optical devices: Volume holograms that are superimposed using angular and peristrophic multiplexing and optical waveguides with a well-defined refractiveindex profile. In the latter, we precisely control the propagating modes via tuning the 3D-printed waveguide parameters and report step-index and graded-index core-cladding transitions.

show abstract

Section: Experimental Demonstrations a Volume Hologramsmentioning

confidence: 99%

Direct (3+1)D laser writing of graded-index optical elements

Porté

Dinç

Moughames

et al. 2021

Optica

Self Cite

View full text Add to dashboard Cite

show abstract

“…Zhou et al quantify this advantage for a range of more general topologies, include learning and confirm that photonic NNs can compete with similar NN models run on GPUs in terms of inference accuracy in several present-day benchmark data sets. Furthermore, the concept is size-scalable: neurons are implemented in 2D-planes while connections leverage optical propagation along the third dimension, and 3D makes the physical footprint scale linear when augmenting the number of neurons 6 .…”

Section: Figure 1|mentioning

confidence: 99%

Competitive photonic neural networks

Brunner

2021

Nat. Photonics

Self Cite

View full text Add to dashboard Cite

“…This potential advantage over electronics was realized decades ago [7,8] and is detailed in [9]. Reducing energy deposition into the substrate could enable novel three-dimensional integration [10] to ultimately realize scalable integration of parallel ANNs [11]. Optical ANNs did outperform their electronic counterparts already at a previous stage [12], and their superior scaling has recently again been demonstrated [13].…”

Section: Introductionmentioning

confidence: 99%

A complete, parallel, and autonomous photonic neural network in a semiconductor multimode laser

Porté

Skalli

Haghighi

et al. 2021

Emerging Topics in Artificial Intelligence (ETAI) 2021

Self Cite

View full text Add to dashboard Cite

Neural networks are one of the disruptive computing concepts of our time. However, they fundamentally differ from classical, algorithmic computing. These differences result in equally fundamental, severe and relevant challenges for neural network computing using current computing substrates. Neural networks urge for parallelism across the entire processor and for a co-location of memory and arithmetic, i.e. beyond von Neumann architectures. Parallelism in particular made photonics a highly promising platform, yet until now scalable and integratable concepts are scarce. Here, we demonstrate for the first time how a fully parallel and fully implemented photonic neural network can be realized by spatially multiplexing neurons across the complex optical near-field of a semiconductor multimode laser. Discrete spatial sampling defines ∼90 nodes on the surface of a large-area vertical cavity surface emitting laser that is optically injected with the artificial neural networks input information. Importantly, all neural network connections are realized in hardware, and our processor produces results without pre-or post-processing. Input and output weights are realized via the complex transmission matrix of a multimode fiber and a digital micro-mirror array, respectively. We train the readout weights to perform 2-bit header recognition, a 2-bit XOR logical function and 2-bit digital to analog conversion, and obtain < 0.9 × 10 −3 and 2.9 × 10 −2 error rates for digit recognition and XOR, respectively. Finally, the digital to analog conversion can be realized with a standard deviation of only 5.4 × 10 −2 . Crucially, our proof-of-concept system is scalable to much larger sizes and to bandwidths in excess of 20 GHz.

show abstract

Optical neural networks: The 3D connection

Cited by 27 publications

References 8 publications

Direct (3+1)D laser writing of graded-index optical elements

Direct (3+1)D laser writing of graded-index optical elements

Competitive photonic neural networks

A complete, parallel, and autonomous photonic neural network in a semiconductor multimode laser

Contact Info

Product

Resources

About