Nonlinear processing with linear optics.

Deep neural networks have achieved remarkable breakthroughs by leveraging multiple layers of data processing to extract hidden representations, albeit at the cost of large electronic computing power. To enhance energy efficiency and speed, the optical implementation of neural networks aims to harness the advantages of optical bandwidth and the energy efficiency of optical interconnections. In the absence of low-power optical nonlinearities, the challenge in the implementation of multilayer optical networks lies in realizing multiple optical layers without resorting to electronic components. Here we present a novel framework that uses multiple scattering, and which is capable of synthesizing programmable linear and nonlinear transformations concurrently at low optical power by leveraging the nonlinear relationship between the scattering potential, represented by data, and the scattered field. Theoretical and experimental investigations show that repeating the data by multiple scattering enables nonlinear optical computing with low-power continuous-wave light. Moreover, we empirically find that scaling of this optical framework follows a power law.

Electrowriting of SU-8 Microfibers.

As microfiber-based additive manufacturing (AM) technologies, melt electrowriting (MEW) and solution electrowriting (SEW) have demonstrated efficacy with more biomedically relevant materials. By processing SU-8 resin using MEW and SEW techniques, a material with substantially different mechanical, thermal, and optical properties than that typically processed is introduced. SU-8 polymer is temperature sensitive and requires the devising of a specific heating protocol to be properly processed. Smooth-surfaced microfibers resulted from MEW of SU8 for a short period (from 30 to 90 min), which provides the greatest control and, thus, reproducibility of the printed microfibers. This investigation explores various parameters influencing the electrowriting process, printing conditions, and post-processing to optimize the fabrication of intricate 3D structures. This work demonstrates the controlled generation of straight filaments and complex multi-layered architectures, which were characterized by brightfield, darkfield, and scanning electron microscopy (SEM). This research opens new avenues for the design and development of 3D-printed photonic systems by leveraging the properties of SU-8 after both MEW and SEW processing.

Volume holograms with linear diffraction efficiency relation by (3 + 1)D printing.

We demonstrate the fabrication of volume holograms using two-photon polymerization with dynamic control of light exposure. We refer to our method as (3 + 1)D printing. Volume holograms that are recorded by interfering reference and signal beams have a diffraction efficiency relation that is inversely proportional to the square of the number of superimposed holograms. By using (3 + 1)D printing for fabrication, the refractive index of each voxel is created independently and thus, by digitally filtering the undesired interference terms, the diffraction efficiency is now inversely proportional to the number of multiplexed gratings. We experimentally demonstrated this linear dependence by recording M = 50 volume gratings. To the best of our knowledge, this is the first experimental demonstration of distributed volume holograms that overcome the 1/M2 limit.

Forward-forward training of an optical neural network.

Neural networks (NNs) have demonstrated remarkable capabilities in various tasks, but their computation-intensive nature demands faster and more energy-efficient hardware implementations. Optics-based platforms, using technologies such as silicon photonics and spatial light modulators, offer promising avenues for achieving this goal. However, training multiple programmable layers together with these physical systems poses challenges, as they are difficult to fully characterize and describe with differentiable functions, hindering the use of error backpropagation algorithm. The recently introduced forward-forward algorithm (FFA) eliminates the need for perfect characterization of the physical learning system and shows promise for efficient training with large numbers of programmable parameters. The FFA does not require backpropagating an error signal to update the weights, rather the weights are updated by only sending information in one direction. The local loss function for each set of trainable weights enables low-power analog hardware implementations without resorting to metaheuristic algorithms or reinforcement learning. In this paper, we present an experiment utilizing multimode nonlinear wave propagation in an optical fiber demonstrating the feasibility of the FFA approach using an optical system. The results show that incorporating optical transforms in multilayer NN architectures trained with the FFA can lead to performance improvements, even with a relatively small number of trainable weights. The proposed method offers a new path to the challenge of training optical NNs and provides insights into leveraging physical transformations for enhancing the NN performance.

Photonic waveguide bundles using 3D laser writing and deep neural network image reconstruction.

In recent years, three-dimensional (3D) printing with multi-photon laser writing has become an essential tool for the manufacturing of three-dimensional optical elements. Single-mode optical waveguides are one of the fundamental photonic components, and are the building block for compact multicore fiber bundles, where thousands of single-mode elements are closely packed, acting as individual pixels and delivering the local information to a sensor. In this work, we present the fabrication of polymer rectangular step-index (STIN) optical waveguide bundles in the IP-Dip photoresist, using a commercial 3D printer. Moreover, we reduce the core-to-core spacing of the imaging bundles by means of a deep neural network (DNN) which has been trained with a large synthetic dataset, demonstrating that the scrambling of information due to diffraction and cross-talk between fiber cores can be undone. The DNN-based approach can be adopted in applications such as on-chip platforms and microfluidic systems where accurate imaging from in-situ printed fiber bundles suffer cross-talk. In this respect, we provide a design and fabrication guideline for such scenarios by employing the DNN not only as a post-processing technique but also as a design optimization tool.