arXiv subjects

Thomas Van Vaerenbergh

Publications and source records attributed to Thomas Van Vaerenbergh.

27 records · Page 2Linked to original sources

arXiv2023

NEUROPULS: NEUROmorphic energy-efficient secure accelerators based on Phase change materials aUgmented siLicon photonicS

This special session paper introduces the Horizon Europe NEUROPULS project, which targets the development of secure and energy-efficient RISC-V interfaced neuromorphic accelerators using augmented silicon photonics technology. Our approach aims to develop an augmented silicon photonics platform, an FPGA-powered RISC-V-connected computing platform, and a complete simulation platform to demonstrate the neuromorphic accelerator capabilities. In particular, their main advantages and limitations will be addressed concerning the underpinning technology for each platform. Then, we will discuss three targeted use cases for edge-computing applications: Global National Satellite System (GNSS) anti-jamming, autonomous driving, and anomaly detection in edge devices. Finally, we will address the reliability and security aspects of the stand-alone accelerator implementation and the project use cases.

cs.AR↗

arXiv2023

Special Session: Neuromorphic hardware design and reliability from traditional CMOS to emerging technologies

The field of neuromorphic computing has been rapidly evolving in recent years, with an increasing focus on hardware design and reliability. This special session paper provides an overview of the recent developments in neuromorphic computing, focusing on hardware design and reliability. We first review the traditional CMOS-based approaches to neuromorphic hardware design and identify the challenges related to scalability, latency, and power consumption. We then investigate alternative approaches based on emerging technologies, specifically integrated photonics approaches within the NEUROPULS project. Finally, we examine the impact of device variability and aging on the reliability of neuromorphic hardware and present techniques for mitigating these effects. This review is intended to serve as a valuable resource for researchers and practitioners in neuromorphic computing.

cs.ET↗

arXiv2023

Adjoint optimization of polarization-splitting grating couplers

We have designed a polarization-splitting grating coupler (PSGC) in silicon-oninsulator (SOI) that has 1.2 dB peak loss in numerical simulations, which is the best simulated performance of PSGCs without a bottom reflector to the best of our knowledge. Adjoint method-based shape optimization enables us to explore complex geometries that are intractable with conventional design approaches. Physics-based process-independent knowledge of PSGCs is extracted from the adjoint optimization and can be transferred to other platforms with a minimum of effort.

physics.optics↗

arXiv2021

Inverse Design of Grating Couplers Using the Policy Gradient Method from Reinforcement Learning

We present a proof-of-concept technique for the inverse design of electromagnetic devices motivated by the policy gradient method in reinforcement learning, named PHORCED (PHotonic Optimization using REINFORCE Criteria for Enhanced Design). This technique uses a probabilistic generative neural network interfaced with an electromagnetic solver to assist in the design of photonic devices, such as grating couplers. We show that PHORCED obtains better performing grating coupler designs than local gradient-based inverse design via the adjoint method, while potentially providing faster convergence over competing state-of-the-art generative methods. As a further example of the benefits of this method, we implement transfer learning with PHORCED, demonstrating that a neural network trained to optimize 8$^\circ$ grating couplers can then be re-trained on grating couplers with alternate scattering angles while requiring >10$\times$ fewer simulations than control cases.

physics.comp-ph↗

arXiv2021

The Boundary for Quantum Advantage in Gaussian Boson Sampling

Identifying the boundary beyond which quantum machines provide a computational advantage over their classical counterparts is a crucial step in charting their usefulness. Gaussian Boson Sampling (GBS), in which photons are measured from a highly entangled Gaussian state, is a leading approach in pursuing quantum advantage. State-of-the-art quantum photonics experiments that, once programmed, run in minutes, would require 600 million years to simulate using the best pre-existing classical algorithms. Here, we present substantially faster classical GBS simulation methods, including speed and accuracy improvements to the calculation of loop hafnians, the matrix function at the heart of GBS. We test these on a $\sim \! 100,000$ core supercomputer to emulate a range of different GBS experiments with up to 100 modes and up to 92 photons. This reduces the run-time of classically simulating state-of-the-art GBS experiments to several months -- a nine orders of magnitude improvement over previous estimates. Finally, we introduce a distribution that is efficient to sample from classically and that passes a variety of GBS validation methods, providing an important adversary for future experiments to test against.

quant-ph↗

arXiv2021

Order-of-magnitude differences in computational performance of analog Ising machines induced by the choice of nonlinearity

Ising machines based on nonlinear analog systems are a promising method to accelerate computation of NP-hard optimization problems. Yet, their analog nature is also causing amplitude inhomogeneity which can deteriorate the ability to find optimal solutions. Here, we investigate how the system's nonlinear transfer function can mitigate amplitude inhomogeneity and improve computational performance. By simulating Ising machines with polynomial, periodic, sigmoid and clipped transfer functions and benchmarking them with MaxCut optimization problems, we find the choice of transfer function to have a significant influence on the calculation time and solution quality. For periodic, sigmoid and clipped transfer functions, we report order-of-magnitude improvements in the time-to-solution compared to conventional polynomial models, which we link to the suppression of amplitude inhomogeneity induced by saturation of the transfer function. This provides insights into the suitability of systems for building Ising machines and presents an efficient way for overcoming performance limitations.

physics.app-ph↗

arXiv2019

Harnessing Intrinsic Noise in Memristor Hopfield Neural Networks for Combinatorial Optimization

We describe a hybrid analog-digital computing approach to solve important combinatorial optimization problems that leverages memristors (two-terminal nonvolatile memories). While previous memristor accelerators have had to minimize analog noise effects, we show that our optimization solver harnesses such noise as a computing resource. Here we describe a memristor-Hopfield Neural Network (mem-HNN) with massively parallel operations performed in a dense crossbar array. We provide experimental demonstrations solving NP-hard max-cut problems directly in analog crossbar arrays, and supplement this with experimentally-grounded simulations to explore scalability with problem size, providing the success probabilities, time and energy to solution, and interactions with intrinsic analog noise. Compared to fully digital approaches, and present-day quantum and optical accelerators, we forecast the mem-HNN to have over four orders of magnitude higher solution throughput per power consumption. This suggests substantially improved performance and scalability compared to current quantum annealing approaches, while operating at room temperature and taking advantage of existing CMOS technology augmented with emerging analog non-volatile memristors.

cs.ET↗

arXiv2018

Thermally tunable hybrid photonic architecture for nonlinear optical circuits

We develop a thermally tunable hybrid photonic platform comprising gallium arsenide (GaAs) photonic crystal cavities, silicon nitride (SiN$_x$) grating couplers and waveguides, and chromium (Cr) microheaters on an integrated photonic chip. The GaAs photonic crystal cavities are evanescently connected to a common bus waveguide, separating the computation and communication layers. The microheaters are designed to continuously and reversibly tune distant photonic crystal cavities to a common resonance. This architecture can be implemented in a coherent optical network for dedicated optical computing and machine learning.

physics.app-ph↗

arXiv2015

Towards Trainable Media: Using Waves for Neural Network-Style Training

In this paper we study the concept of using the interaction between waves and a trainable medium in order to construct a matrix-vector multiplier. In particular we study such a device in the context of the backpropagation algorithm, which is commonly used for training neural networks. Here, the weights of the connections between neurons are trained by multiplying a `forward' signal with a backwards propagating `error' signal. We show that this concept can be extended to trainable media, where the gradient for the local wave number is given by multiplying signal waves and error waves. We provide a numerical example of such a system with waves traveling freely in a trainable medium, and we discuss a potential way to build such a device in an integrated photonics chip.

cs.NE↗