arXiv subjects

Xavier Porte

Publications and source records attributed to Xavier Porte.

26 records · Page 2Linked to original sources

arXiv2020

Three dimensional waveguide-interconnects for scalable integration of photonic neural networks

Photonic waveguides are prime candidates for integrated and parallel photonic interconnects. Such interconnects correspond to large-scale vector matrix products, which are at the heart of neural network computation. However, parallel interconnect circuits realized in two dimensions, for example by lithography, are strongly limited in size due to disadvantageous scaling. We use three dimensional (3D) printed photonic waveguides to overcome this limitation. 3D optical-couplers with fractal topology efficiently connect large numbers of input and output channels, and we show that the substrate's footprint area scales linearly. Going beyond simple couplers, we introduce functional circuits for discrete spatial filters identical to those used in deep convolutional neural networks.

cs.ET↗

arXiv2019

Reservoir-size dependent learning in analogue neural networks

The implementation of artificial neural networks in hardware substrates is a major interdisciplinary enterprise. Well suited candidates for physical implementations must combine nonlinear neurons with dedicated and efficient hardware solutions for both connectivity and training. Reservoir computing addresses the problems related with the network connectivity and training in an elegant and efficient way. However, important questions regarding impact of reservoir size and learning routines on the convergence-speed during learning remain unaddressed. Here, we study in detail the learning process of a recently demonstrated photonic neural network based on a reservoir. We use a greedy algorithm to train our neural network for the task of chaotic signals prediction and analyze the learning-error landscape. Our results unveil fundamental properties of the system's optimization hyperspace. Particularly, we determine the convergence speed of learning as a function of reservoir size and find exceptional, close to linear scaling. This linear dependence, together with our parallel diffractive coupling, represent optimal scaling conditions for our photonic neural network scheme.

cs.NE↗

arXiv2019

Fundamental aspects of noise in analog-hardware neural networks

We study and analyze the fundamental aspects of noise propagation in recurrent as well as deep, multi-layer networks. The main focus of our study are neural networks in analogue hardware, yet the methodology provides insight for networks in general. The system under study consists of noisy linear nodes, and we investigate the signal-to-noise ratio at the network's outputs which is the upper limit to such a system's computing accuracy. We consider additive and multiplicative noise which can be purely local as well as correlated across populations of neurons. This covers the chief internal-perturbations of hardware networks and noise amplitudes were obtained from a physically implemented recurrent neural network and therefore correspond to a real-world system. Analytic solutions agree exceptionally well with numerical data, enabling clear identification of the most critical components and aspects for noise management. Focusing on linear nodes isolates the impact of network connections and allows us to derive strategies for mitigating noise. Our work is the starting point in addressing this aspect of analogue neural networks, and our results identify notoriously sensitive points while simultaneously highlighting the robustness of such computational systems.

cs.ET↗

arXiv2019

Diffractive coupling for photonic networks: how big can we go?

Photonic networks are considered a promising substrate for high-performance future computing systems. Compared to electronics, photonics has significant advantages for a fully parallel implementation of networks. A promising approach for parallel large-scale photonic networks is realizing the connections using diffraction. Here, we characterize the scalability of such diffractive coupling in great detail. Based on experiments, analytically obtained bounds and numerical simulations considering real-world optical imaging setups, we find that the concept in principle enables networks hosting over a million optical emitters. This would be a breakthrough in multiple areas, illustrating a clear path toward future, large scale photonic networks.

physics.optics↗

arXiv2019

Coupled nonlinear delay systems as deep convolutional neural networks

Neural networks are currently transforming the field of computer algorithms, yet their emulation on current computing substrates is highly inefficient. Reservoir computing was successfully implemented on a large variety of substrates and gave new insight in overcoming this implementation bottleneck. Despite its success, the approach lags behind the state of the art in deep learning. We therefore extend time-delay reservoirs to deep networks and demonstrate that these conceptually correspond to deep convolutional neural networks. Convolution is intrinsically realized on a substrate level by generic drive-response properties of dynamical systems. The resulting novelty is avoiding vector-matrix products between layers, which cause low efficiency in today's substrates. Compared to singleton time-delay reservoirs, our deep network achieves accuracy improvements by at least an order of magnitude in Mackey-Glass and Lorenz timeseries prediction.

cs.ET↗

arXiv2018

Mutual coupling and synchronization of optically coupled quantum-dot micropillar lasers at ultra-low light levels

In this work we explore the limits of synchronization of mutually coupled oscillators at the crossroads of classical and quantum physics. In order to address this uncovered regime of synchronization we apply electrically driven quantum dot micropillar lasers operating in the regime of cavity quantum electrodynamics. These high-$β$ microscale lasers feature cavity enhanced coupling of spontaneous emission and operate at output powers on the order of 100 nW. We selected pairs of micropillar lasers with almost identical optical properties in terms of the input-output dependence and the emission energy which we mutually couple over a distance of about 1m and bring into spectral resonance by precise temperature tuning. By excitation power and detuning dependent studies we unambiguously identify synchronization of two mutually coupled high-$β$ microlasers via frequency locking associated with a sub-GHz locking range. A detailed analysis of the synchronization behavior includes theoretical modeling based on semi-classical stochastic rate equations and reveals striking differences from optical synchronization in the classical domain with negligible spontaneous emission noise and optical powers usually well above the mW range. In particular, we observe deviations from the classically expected locking slope and broadened locking boundaries which are successfully explained by the fact the quantum noise plays an important role in our cavity enhanced optical oscillators. Beyond that, introducing additional self-feedback to the two mutually coupled microlasers allows us to realize zero-lag synchronization. Our work provides important insight into synchronization of optical oscillators at ultra-low light levels and has high potential to pave the way for future experiments in the quantum regime of synchronization.

cond-mat.mes-hall↗

arXiv2017

Quantum-optical spectroscopy of a two-level system using an electrically driven micropillar laser as resonant excitation source

Two-level emitters constitute main building blocks of photonic quantum systems and are model systems for the exploration of quantum optics in the solid state. Most interesting is the strict-resonant excitation of such emitters to generate close to ideal quantum light and to control their occupation coherently. Up till now related experiments have been performed exclusively using bulky lasers which hinders the application of resonantly driven two-level emitters in photonic quantum systems. Here we perform quantum-optical spectroscopy of a two-level system using a compact high-$β$ microlaser as excitation source. The two-level system is based on a semiconductor quantum dot (QD), which is excited resonantly by a fiber-coupled electrically driven micropillar laser. In this way we dress the excitonic state of the QD under continuous wave excitation and trigger the emission of single-photons with strong multi-photon suppression ($g^{(2)}(0)=0.02$) and high photon indistinguishably ($V=57\pm9\%$) via pulsed resonant excitation at 156 MHz.

cond-mat.mes-hall↗

arXiv2017

Micropillars with a controlled number of site-controlled quantum dots

We report on the realization of micropillars with site-controlled quantum dots (SCQDs) in the active layer. The SCQDs are grown via the buried stressor approach which allows for the positioned growth and device integration of a controllable number of QDs with high optical quality. This concept is very powerful as the number and the position of SCQDs in the cavity can be simultaneously controlled by the design of the buried stressor. The fabricated micropillars exhibit a high degree of position control for the QDs above the buried stressor and $Q$-factors of up to 12000 at an emission wavelength around 930 nm. We experimentally analyze and numerically model the cavity $Q$-factor, the mode volume, the Purcell factor and the photon-extraction efficiency as a function of the aperture diameter of the buried stressor. Exploiting these SCQD micropillars, we experimentally observe the Purcell enhancement in the single-QD regime with $F_P$ = 4.3 $\pm$ 0.3.

cond-mat.mes-hall↗