A5007928811- Photonic and Optical Devices
- Neural Networks and Reservoir Computing
- Optical Network Technologies
- Advanced Memory and Neural Computing
- Photonic Crystals and Applications
- Advanced Photonic Communication Systems
- Quantum Computing Algorithms and Architecture
- Optical Coatings and Gratings
- Advanced Fiber Laser Technologies
- Ferroelectric and Negative Capacitance Devices
- Semiconductor Lasers and Optical Devices
- Quantum Information and Cryptography
- Neural Networks and Applications
- Neural dynamics and brain function
- Parallel Computing and Optimization Techniques
- Mechanical and Optical Resonators
- Network Packet Processing and Optimization
- Advanced Fiber Optic Sensors
- Plasmonic and Surface Plasmon Research
- Cellular Automata and Applications
- Integrated Circuits and Semiconductor Failure Analysis
- Advanced Optical Imaging Technologies
- Quantum-Dot Cellular Automata
- Stochastic Gradient Optimization Techniques
- Neural Networks Stability and Synchronization
Hewlett Packard Enterprise (United States)
2016-2025
Hewlett-Packard (United States)
2016-2025
Ghent University
2011-2023
University of California, Berkeley
2020
Cell Gate (United States)
2015-2016
Photonics (United States)
2014
Integrated Optoelectronics (Norway)
2012
Abstract An overview is presented of the current state‐of‐the‐art in silicon nanophotonic ring resonators. Basic theory resonators discussed, and applied to peculiarities submicron photonic wire waveguides: small dimensions tight bend radii, sensitivity perturbations boundary conditions fabrication processes. Theory compared quantitative measurements. Finally, several more promising applications are discussed: filters optical delay lines, label‐free biosensors, active rings for efficient...
In today's age, companies employ machine learning to extract information from large quantities of data. One those techniques, reservoir computing (RC), is a decade old and has achieved state-of-the-art performance for processing sequential Dedicated hardware realizations RC could enable speed gains power savings. Here we propose the first integrated passive silicon photonics reservoir. We demonstrate experimentally through simulations that, thanks paradigm, this generic chip can be used...
To emulate a spiking neuron, photonic component needs to be excitable. In this paper, we theoretically simulate and experimentally demonstrate cascadable excitability near self-pulsation regime in high-Q-factor silicon-on-insulator microrings. For the theoretical study use Coupled Mode Theory. While neglecting fast energy phase dynamics of cavity light, can still preserve most important microring dynamics, by only keeping temperature difference with surroundings amount free carriers as...
Abstract Silicon microring resonators very often exhibit resonance splitting due to backscattering. This effect is hard quantitatively and predicatively model. paper presents a behavioral circuit model for microrings that explains the wide variations in observed experiments. The based on an in‐depth analysis of contributions backscattering by both waveguides couplers. Backscattering transforms unidirectional into bidirectional circuits coupling clockwise counterclockwise circulating modes....
A detailed analysis of fundamental tradeoffs between ring radius and coupling gap size is presented to draw realistic borders the possible design space for microring resonators (MRRs). The coefficient ring-waveguide structure estimated based on an integration nonuniform waveguide. Combined with supermode two coupled waveguides, this approach further expanded into a closed-form equation that describes strength. This permits evaluate how distance separating waveguide from resonator, radius,...
Identifying the boundary beyond which quantum machines provide a computational advantage over their classical counterparts is crucial step in charting usefulness. Gaussian boson sampling (GBS), photons are measured from highly entangled state, leading approach pursuing advantage. State-of-the-art GBS experiments that run minutes would require 600 million years to simulate using best preexisting algorithms. Here, we present faster simulation methods, including speed and accuracy improvements...
Heterogeneous III-V-on-silicon photonic integration has proved to be an attractive and volume manufacturable solution that marries the merits of III-V compounds silicon technology for various integrated circuit (PIC) applications. The current main-stream Ethernet trends larger bandwidth are pushing higher modulation baudrate or employing advanced format datacom However, neither is likely able significantly drive overall cost energy efficiency best sweet spot, nor unfold full potential...
The convergence of deep learning and big data has spurred significant interest in developing novel hardware that can run large artificial intelligence (AI) workloads more efficiently. Over the last several years, silicon photonics emerged as a disruptive technology for next-generation accelerators machine (ML). More recently, heterogeneous integration III-V compound semiconductors opened door to integrating lasers semiconductor optical amplifiers at wafer-scale, enabling scaling size,...
We present a tool that aids in the modeling of optical circuits, both frequency and time domain. The is based on definition node, which can have an instantaneous input-output relation different state variables (e.g., temperature carrier density) differential equations for these states. Furthermore, each node has access to part its input history, allowing creation delay lines or digital filters. Additionally, contain subnodes, hierarchical networks. This be used numerous applications such as...
We demonstrate class I excitability in optically injected microdisk lasers, and propose a possible optical spiking neuron design.The has clear threshold an integrating behavior, leading to output rate-input rate dependency that is comparable the characteristic of sigmoidal artificial neurons.We also show phase input pulses influence on response, can be used create inhibitory, as well excitatory perturbations.
This paper proposes a large-scale, energy-efficient, high-throughput, and compact tensorized optical neural network (TONN) exploiting the tensor-train decomposition architecture on an integrated III–V-on-silicon metal–oxide–semiconductor capacitor (MOSCAP) platform. The proposed TONN is scalable to 1024 × synapses beyond, which extremely difficult for conventional ONN architectures by using cascaded multi-wavelength small-radix (e.g., 8 8) tensor cores. Simulation experiments show that uses...
Specialized function gradient computing hardware could greatly improve the performance of state-of-the-art optimization algorithms. Prior work on such hardware, performed in context Ising Machines and related concepts, is limited to quadratic polynomials not scalable commonly used higher-order functions. Here, we propose an approach for massively parallel calculations high-degree polynomials, which conducive efficient mixed-signal in-memory circuit implementations whose area scales...
Machine learning algorithms, and more in particular neural networks, arguably experience a revolution terms of performance. Currently, the best systems we have for speech recognition, computer vision similar problems are based on trained using half-century old backpropagation algorithm. Despite fact that networks form analog computers, they still implemented digitally reasons convenience availability. In this paper demonstrate how can design physical linear dynamic with non-linear feedback...
We propose and demonstrate an accurate method of measuring the effective refractive index thermo-optic coefficient silicon-on-insulator waveguides in entire C-band using three Mach-Zehnder interferometers. The allows for extraction wavelength dispersion takes into account fabrication variability. Wafer scale measurements are performed variations presented different waveguide widths: 450, 600, 800 nm, TE polarization. is generic can be applied to other geometries material systems wavelengths...
Reservoir computing (RC) is a technique in machine learning inspired by neural systems. RC has been used successfully to solve complex problems such as signal classification and generation. These systems are mainly implemented software, thereby they limited speed power efficiency. Several optical optoelectronic implementations have demonstrated, which the system signals with an amplitude phase. It proven that these enrich dynamics of system, beneficial for performance. In this paper, we...
Silicon-on-insulator microrings both self-pulsate and are excitable due to the presence of thermal free-carrier-related nonlinearities. We show how a dimensionless mean-field model, in which fast light dynamics neglected only temperature amount free carriers remain as variables, can explain this dynamic behavior. Apart from scaled detuning input wavelength resonance power, system contains limited number parameters dependent on geometry material cavity. Moreover, onset oscillation is still...
We propose an integrated photonic circuit that acts as optical coherent Ising machine and simulates its performance on the basis of some example problems. In contrast to previous all-optical approaches, proposed does not require parametric oscillator can, hence, operate at a single wavelength, reducing overall design complexity. present symmetric nonlinear device fundamental building block can use self-phase modulation in two microring resonators emulate continuously tunable, symmetrically...
We present a proof-of-concept technique for the inverse design of electromagnetic devices motivated by policy gradient method in reinforcement learning, named PHORCED (PHotonic Optimization using REINFORCE Criteria Enhanced Design). This uses probabilistic generative neural network interfaced with an solver to assist photonic devices, such as grating couplers. show that obtains better performing coupler designs than local gradient-based via adjoint method, while potentially providing faster...
Matrix-vector multiplications (MVMs) are essential for a wide range of applications, particularly in modern machine learning and quantum computing. In photonics, there is growing interest developing architectures capable performing linear operations with high speed, low latency, minimal loss. Traditional interferometric photonic architectures, such as the Clements design, have been extensively used MVM operations. However, these scale, improving stability robustness becomes critical. this...
Solving optimization problems is a highly demanding workload requiring high-performance computing systems. Optimization solvers are usually difficult to parallelize in conventional digital architectures, particularly when stochastic decisions involved. Recently, analog architectures for accelerating have been presented, but they were limited academic quadratic polynomial format. Here we present KLIMA, k-Local In-Memory Accelerator with resistive Content Addressable Memories (CAMs) and...
<title>Abstract</title> The need for high-speed, energy-efficient computing in machine learning and real-time communication necessitates innovations beyond conventional digital analog electronics to sustain computational power advances without requiring prohibitive energy amounts. Photonics has emerged as a promising platform demonstrating significant highlights the field of linear transformations. Adopting, however, use photons within broad range applications their successful employment...