Optical Neural Network Architecture for Deep Learning with Temporal Synthetic Dimension

The physical concept of synthetic dimensions has recently been introduced into optics.The fundamental physics and applications are not yet fully understood,and this report explores an approach to optical neural networks using synthetic dimension in time domain,by theoretically proposing to utilize a single resonator network,where the arrival times of optical pulses are interconnected to construct a temporal synthetic dimension.The set of pulses in each roundtrip therefore provides the sites in each layer in the optical neural network,and can be linearly transformed with splitters and delay lines,including the phase modulators,when pulses circulate inside the network.Such linear transformation can be arbitrarily controlled by applied modulation phases,which serve as the building block of the neural network together with a nonlinear component for pulses.We validate the functionality of the proposed optical neural network for the deep learning purpose with examples handwritten digit recognition and optical pulse train distribution classification problems.This proof of principle computational work explores the new concept of developing a photonics-based machine learning in a single ring network using synthetic dimensions,which allows flexibility and easiness of reconfiguration with complex functionality in achieving desired optical tasks.

Multi-dimensional band structure spectroscopy in the synthetic frequency dimension

The concept of synthetic dimensions in photonics provides a versatile platform in exploring multi-dimensional physics.Many of these physics are characterized by band structures in more than one dimensions.Existing efforts on band structure measurements in the photonic synthetic frequency dimension however are limited to either onedimensional Brillouin zones or one-dimensional subsets of multi-dimensional Billouin zones.Here we theoretically propose and experimentally demonstrate a method to fully measure multi-dimensional band structures in the synthetic frequency dimension.We use a single photonic resonator under dynamical modulation to create a multidimensional synthetic frequency lattice.We show that the band structure of such a lattice over the entire multidimensional Brillouin zone can be measured by introducing a gauge potential into the lattice Hamiltonian.Using this method,we perform experimental measurements of two-dimensional band structures of a Hermitian and a non-Hermitian Hamiltonian.The measurements reveal some of the general properties of point-gap topology of the non-Hermitian Hamiltonian in more than one dimensions.Our results demonstrate experimental capabilities to fully characterize high-dimensional physical phenomena in the photonic synthetic frequency dimension.

A wireless, implantable optoelectrochemical probe for optogenetic stimulation and dopamine detection

Physical and chemical technologies have been continuously progressing advances in neuroscience research.The development of research tools for closed-loop control and monitoring neural activities in behaving animals is highly desirable.In this paper,we introduce a wirelessly operated,miniaturized microprobe system for optical interrogation and neurochemical sensing in the deep brain.Via epitaxial liftoff and transfer printing,microscale light-emitting diodes(micro-LEDs)as light sources and poly(3,4-ethylenedioxythiophene)polystyrene sulfonate(PEDOT:PSS)-coated diamond films as electrochemical sensors are vertically assembled to form implantable optoelectrochemical probes for real-time optogenetic stimulation and dopamine detection capabilities.A customized,lightweight circuit module is employed for untethered,remote signal control,and data acquisition.After the probe is injected into the ventral tegmental area(VTA)of freely behaving mice,in vivo experiments clearly demonstrate the utilities of the multifunctional optoelectrochemical microprobe system for optogenetic interference of place preferences and detection of dopamine release.The presented options for material and device integrations provide a practical route to simultaneous optical control and electrochemical sensing of complex nervous systems.

Optoelectronic sensing of biophysical and biochemical signals based on photon recycling of a micro-LED

Conventional bioelectrical sensors and systems integrate multiple power harvesting,signal amplification and data transmission components for wireless biological signal detection.This paper reports the real-time biophysical and biochemical activities can be optically captured using a microscale light-emitting diode(micro-LED),eliminating the need for complicated sensing circuit.Such a thin-film diode based device simultaneously absorbs and emits photons,enabling wireless power harvesting and signal transmission.Additionally,owing to its strong photon-recycling effects,the micro-LED^photoluminescence(PL)emission exhibits a superlinear dependence on the external conductance.Taking advantage of these unique mechanisms,instantaneous biophysical signals including galvanic skin response,pressure and temperature,and biochemical signals like ascorbic acid concentration,can be optically monitored,and it demonstrates that such an optoelectronic sensing technique outperforms a traditional tethered,electrically based sensing circuit,in terms of its footprint,accuracy and sensitivity.This presented optoelectronic sensing approach could establish promising routes to advanced biological sensors.

Ultrafast and low-power optoelectronic infrared-to-visible upconversion devices

Photon upconversion with transformation of low-energy photons to high-energy photons has been widely studied and especially applied in biomedicine for sensing, stimulation, and imaging. Conventional upconversion materials rely on nonlinear luminescence processes, suffering from long decay lifetime or high excitation power. Here,we present a microscale, optoelectronic infrared-to-visible upconversion device design that can be excited at low power(1–100 mW∕cm^2). By manipulating device geometry, illumination position, and temperature, the device luminescence decay lifetime can be tuned from tens to hundreds of nanoseconds. Based on carrier transportation and circuit dynamics, theoretical models are established to understand the transient behaviors. Compared with other mechanisms, the optoelectronic upconversion approach demonstrates the shortest luminescence lifetime with the lowest required excitation power, owing to its unique photon–electron conversion process. These features are expected to empower the device with essential capabilities for versatile applications as high-performance light emitters.