Reservoir computing is a computational framework suited for temporal/sequential data processing. It is derived from several recurrent neural network models, including echo state networks and liquid state machines. A reservoir computing system consists of a reservoir for mapping inputs into a high-dimensional space and a readout for pattern analysis from the high-dimensional states in the reservoir. The reservoir is fixed and only the readout is trained with a simple method such as linear regression and classification. Thus, the major advantage of reservoir computing compared to other recurrent neural networks is fast learning, resulting in low training cost. Another advantage is that the reservoir without adaptive updating is amenable to hardware implementation using a variety of physical systems, substrates, and devices. In fact, such physical reservoir computing has attracted increasing attention in diverse fields of research. The purpose of this review is to provide an overview of recent advances in physical reservoir computing by classifying them according to the type of the reservoir. We discuss the current issues and perspectives related to physical reservoir computing, in order to further expand its practical applications and develop next-generation machine learning systems.
InGaAsN:Sb/GaAs quantum wells (QWs) were grown by solid-source molecular beam epitaxy using a N2 radio-frequency plasma source. Photoluminescence reveals an enhancement in the optical properties of InGaAsN/GaAs QWs by the introduction of Sb flux during growth. X-ray diffraction and reflection high-energy electron diffraction analyses indicate that Sb acts as a surfactant. This technique was used to improve the performance of long-wavelength InGaAsN laser diodes. A low-threshold current density of 520 A/cm2 was achieved for an InGaAsN:Sb/GaAs single quantum well 1.2 μm laser diode at room temperature under pulsed operation.
We demonstrate a GaAs-based p-i-n resonant-cavity-enhanced (RCE) GaInNAs photodetector operating near 1.3 μm. The device design was optimized using a transfer matrix method and experimental absorption spectra obtained from p-i-n structures grown without a resonant cavity. The RCE photodetector was fabricated in a single growth step by using GaAs/AlAs distributed Bragg reflectors for the top and bottom mirrors. A 72% quantum efficiency was obtained with a full width at half maximum of 11 nm.
1.3 μm InGaAsN:Sb/GaAs multiple-quantum-well laser diodes have been grown by solid-source molecular-beam epitaxy using Sb as a surfactant. A low threshold of 1.1 kA/cm2 was achieved for broad-area laser diodes under pulsed operation at room temperature. High-temperature device characterization revealed characteristic temperatures (T0) of 92 and 54 K for operating temperatures below and above 75 °C, respectively, as well as a lasing-wavelength temperature dependence of 0.36 nm/ °C.
Articles you may be interested inCharacteristic of rapid thermal annealing on Ga In ( N ) ( Sb ) As ∕ Ga As quantum well grown by molecular-beam epitaxy J. Appl. Phys. 99, 034903 (2006); 10.1063/1.2164539Thermal excitation effects of photoluminescence of annealed Ga In N As ∕ Ga As quantum-well laser structures grown by plasma-assisted molecular-beam epitaxy Thermally induced diffusion in Ga In N As ∕ Ga As and Ga In As ∕ Ga As quantum wells grown by solid source molecular beam epitaxy
The effect of a variation of the indium and nitrogen concentrations in InxGa1−xAs1−yNy/GaAs multiquantum wells grown by molecular beam epitaxy is studied systematically by room temperature photoreflectance spectroscopy. The band gap redshift caused by a nitrogen fraction of 1.5% decreases by as much as 30% as the indium fraction increases from 0% to 20%. A moderate increase of electron effective mass (Δme∼0.03 m0) is found in all samples containing nitrogen (y≳1%). In compressively strained quantum wells, the energy separation between the first confined heavy and light hole energy levels decreases in a regular manner as the nitrogen fraction increases from 0% to 1.7%, suggesting that the modification of the valence bands due to nitrogen incorporation can be explained by the strain variation.
A CO2 laser-based system was used to provoke the vapor-assisted removal of contaminating particles from different kinds of surfaces. Particles of alumina, silicon carbide, boron carbide, and cerium dioxide, with a size as small as 0.1 μm, have been efficiently removed from silicon, gold, and silicon dioxide surfaces. The dependence of the cleaning efficiency on the laser fluence was investigated; a threshold was found at 0.65 J/cm2 and the efficiency was highest for a fluence ranging from 2.9 to 3.2 J/cm2 for silicon, and 3.2 J/cm2 for gold and silicon dioxide surfaces. The amount of the water vapor which condenses at the surface was also found to play a major role, the best results being obtained with a condensed thickness calculated to be 6 μm. The zeta potential value of the contaminant particles with respect to that of the surface greatly influences the cleaning process.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.