arXiv subjects

Shuhei Ohno

Publications and source records attributed to Shuhei Ohno.

4 recordsLinked to original sources

arXiv2024

Symmetric silicon microring resonator optical crossbar array for accelerated inference and training in deep learning

Photonic integrated circuits are emerging as a promising platform for accelerating matrix multiplications in deep learning, leveraging the inherent parallel nature of light. Although various schemes have been proposed and demonstrated to realize such photonic matrix accelerators, the in-situ training of artificial neural networks using photonic accelerators remains challenging due to the difficulty of direct on-chip backpropagation on a photonic chip. In this work, we propose a silicon microring resonator (MRR) optical crossbar array with a symmetric structure that allows for simple on-chip backpropagation, potentially enabling the acceleration of both the inference and training phases of deep learning. We demonstrate a $4 \times 4$ circuit on a Si-on-insulator (SOI) platform and use it to perform inference tasks of a simple neural network for classifying Iris flowers, achieving a classification accuracy of 93.3%. Subsequently, we train the neural network using simulated on-chip backpropagation and achieve an accuracy of 91.1% in the same inference task after training. Furthermore, we simulate a convolutional neural network (CNN) for handwritten digit recognition, using a $9 \times 9$ MRR crossbar array to perform the convolution operations. This work contributes to the realization of compact and energy-efficient photonic accelerators for deep learning.

cs.ET↗

arXiv2024

Structure of Heavy Mesons in the Light-Front Quark Model

We investigate the structure of ground-state heavy mesons within the light-front quark model, utilizing wave functions derived from the Single Gaussian Ansatz (SGA) and the Gaussian Expansion Method (GEM). By performing a $\chi^2$ fit to static properties such as mass spectra and decay constants, we determine the model parameters for each approach. We then compare the impacts of both methods on the light-front wave functions and structural observables. Our analysis reveals significant differences in the distribution amplitudes (DAs) $\phi_{2;M}(x)$ near the endpoints, with GEM showing enhanced amplitudes and correct asymptotic behavior $\phi_{2;M}(x \to 1) \propto (1-x)$, consistent with perturbative QCD. This endpoint behavior is linked to the short-range (high-momentum) wave function governed by color Coulomb interaction and relativistic kinematics. GEM accurately reproduces a power-law damping $\psi_0(k \to \infty) \propto 1/k_\perp^2$, aligning with perturbative QCD predictions. Furthermore, the electromagnetic form factors of pseudoscalar mesons in the low-$Q^2$ region fall off faster with GEM than with SGA. Overall, while both methods adequately describe static properties, GEM provides a more accurate description of structural properties, being more sensitive to details and asymptotic behaviors.

hep-ph↗

arXiv2021

Ultrahigh-sensitivity optical power monitor for Si photonic circuits

A phototransistor is a promising candidate as an optical power monitor in Si photonic circuits since the internal gain of photocurrent enables high sensitivity. However, state-of-the-art waveguide-coupled phototransistors suffer from a responsivity of lower than $10^3$ A/W, which is insufficient for detecting very low power light. Here, we present a waveguide-coupled phototransistor consisting of an InGaAs ultrathin channel on a Si waveguide working as a gate electrode to increase the responsivity. The Si waveguide gate underneath the InGaAs ultrathin channel enables the effective control of transistor current without optical absorption by the gate metal. As a result, our phototransistor achieved the highest responsivity of approximately $10^6$ A/W among the waveguide-coupled phototransistors, allowing us to detect light of 621 fW propagating in the Si waveguide. The high responsivity and the reasonable response time of approximately 100 $μ$s make our phototransistor promising as an effective optical power monitor in Si photonics circuits.

physics.app-ph↗

arXiv2021

Si microring resonator crossbar array for on-chip inference and training of optical neural network

Deep learning is one of the most advancing technologies in various fields. Facing the limits of the current electronics platform, optical neural networks (ONNs) based on Si programmable photonic integrated circuits (PICs) have attracted considerable attention as a novel deep learning scheme with optical-domain matrix-vector multiplication (MVM). However, most of the proposed Si programmable PICs for ONNs have several drawbacks such as low scalability, high power consumption, and lack of frameworks for training. To address these issues, we have proposed a microring resonator (MRR) crossbar array as a Si programmable PIC for an ONN. In this article, we present a prototype of a fully integrated 4 ${\rm \times}$ 4 MRR crossbar array and demonstrated a simple MVM and classification task. Moreover, we propose on-chip backpropagation using the transpose matrix operation of the MRR crossbar array, enabling the on-chip training of the ONN. The proposed ONN scheme can establish a scalable, power-efficient deep learning accelerator for applications in both inference and training tasks.

cs.ET↗