Ensemble learning of diffractive optical networks

A plethora of research advances have emerged in the fields of optics and photonics that benefit from harnessing the power of machine learning.Specifically,there has been a revival of interest in optical computing hardware due to its potential advantages for machine learning tasks in terms of parallelization,power efficiency and computation speed.Diffractive deep neural networks(D²NNs)form such an optical computing framework that benefits from deep learning-based design of successive diffractive layers to all-optically process information as the input light diffracts through these passive layers.D²NNs have demonstrated success in various tasks,including object classification,the spectral encoding of information,optical pulse shaping and imaging.Here,we substantially improve the inference performance of diffractive optical networks using feature engineering and ensemble learning.After independently training 1252 D²NNs that were diversely engineered with a variety of passive input filters,we applied a pruning algorithm to select an optimized ensemble of D²NNs that collectively improved the image classification accuracy.Through this pruning,we numerically demonstrated that ensembles of N=14 and N=30 D²NNs achieve blind testing accuracies of 61.14±0.23%and 62.13±0.05%,respectively,on the classification of GFAR-10 test images,providing an inference improvennent of>16%compared to the average performance of the individual D²NNs within each ensemble.These results constitute the highest inference accuracies achieved to date by any diffractive optical neural network design on the same dataset and might provide a significant leap to extend the application space of diffractive optical image classification and machine vision systems.     

Extended sliding frame R-Aloha: Medium access control (MAC) protocol for mobile networks

Proliferation of mobile communication devices necessitates a reliable and efficient medium access control (MAC) protocol. In this paper, A MAC protocol, called extended sliding frame reservation Aloha (ESFRA), based on sliding frame R-Aloha (SFRA) is proposed for network access technique. ESFRA is particularly designed to solve the mobile hidden station (MHS) problem in a mobile ad hoc network (MANET) by including relative locations of transmitting stations in the packet frame information header. The MHS problem is unique in mobile networks and occurs if a mobile station enters in a collision free zone of any ongoing communication and disturbs this communication with its transmission. In addition to the MHS problem, ESFRA simultaneously solves hidden station, exposed station, and neighborhood capture problems typically observed in wireless networks. A Markov model of ESFRA is developed and provided here to estimate throughput, delay and collision probabilities of the proposed protocol. The Markov modeling is extended to the analysis of SFRA and IEEE 802.11 to compare these competing MAC protocols with ESFRA. The analysis shows that ESFRA decreases frame transmission delay, increases throughput, and reduces collision probabilities compared to IEEE 802.11 and SFRA. ESFRA improves the network throughput 28 percent compared to that of IEEE 802.11, and 33 percent compared to that of SFRA. The improved performance is obtained at the expense of the synchronization compared to IEEE 802.11, but there is virtually no extra cost compared to SFRA.     

PERFORMANCE OF THREE STAGE TURBO－EQUALIZATION－DECODING

An increasing demand for high data rate transmission and protection over bandlimited channels with severe inter-symbol interference has resulted in a flurry of activity to improve channel equalization. In conjunction with equalization, channel coding-decoding can be employed to improve system performance. In this letter, the performance of the three stageturbo-equalization-decoding employing log maximum a posteriori probability is experimentally evaluated by a fading simulator. The BER is evaluated using various information sequence and interleaver sizes taking into account that the communication medium is a noisy inter symbol interference channel.     

Spin Filtering through Single‐Wall Carbon Nanotubes Functionalized with Single‐Stranded DNA

High spin polarization materials or spin filters are key components in spintronics, a niche subfield of electronics where carrier spins play a functional role. Carrier transmission through these materials is “spin selective,” that is, these materials are able to discriminate between “up” and “down” spins. Common spin filters include transition metal ferromagnets and their alloys, with typical spin selectivity (or, polarization) of ≈50% or less. Here carrier transport is considered in an archetypical one‐dimensional molecular hybrid in which a single wall carbon nanotube (SWCNT) is wrapped around by single stranded deoxyribonucleic acid (ssDNA). By magnetoresistance measurements it is shown that this system can act as a spin filter with maximum spin polarization approaching ≈74% at low temperatures, significantly larger than transition metals under comparable conditions. Inversion asymmetric helicoidal potential of the charged ssDNA backbone induces a Rashba spin‐orbit interaction in the SWCNT channel and polarizes carrier spins. The results are consistent with recent theoretical work that predicted spin dependent conductance in ssDNA‐SWCNT hybrid. Ability to generate highly spin polarized carriers using molecular functionalization can lead to magnet‐less and contact‐less spintronic devices in the future. This can eliminate the conductivity mismatch problem and open new directions for research in organic spintronics.     

Effect of Gaseous Ozone on Papaya Anthracnose

Anthracnose caused by the fungus Colletotrichum gloeosporioides is considered as one of the most devastating postharvest disease of papaya. The aim of this study was to evaluate the effectiveness of gaseous ozone as a potential antifungal preservation technique to overcome anthracnose disease of papaya during cold storage. Different concentrations of ozone (0 (control), 0.04, 1.6, and 4 ppm) were applied for various exposure durations (48, 96, and 144 h). Radial mycelia growth and conidial germination were evaluated in vitro after fungal exposure to the different levels and durations of ozone. Significant inhibition in radial mycelia growth of C. gloeosporioides was observed (p?<?0.05) in all ozone treatments as compared to the control during 8 days of incubation at room temperature (25?±?3 °C). Ozone treatment of papaya fruit with 1.6-ppm ozone for 96 h delayed and simultaneously decreased the disease incidence to 40 % whereas disease severity was rated at 1.7, following 28 days of storage at 12?±?1 °C and 80 % relative humidity. The scanning electron microscopy showed that 4-ppm ozone caused disintegration of spore structure and did not affect the cuticular surface of fruit. Thus, ozone fumigation can reduce postharvest losses of papaya caused by anthracnose.     

Comparison of electrothermal and hydride generation atomic absorption spectrometry for the determination of total arsenic in broiler chicken

The purpose of this study was to estimate total arsenic concentration in different tissues (leg, breast, liver and heart) of broiler chicken by hydride generation atomic absorption spectrometry (HGAAS) and graphite furnace atomic absorption spectrometry (GFAAS), prior to microwave assisted acid digestion. The accuracy of the techniques was evaluated by using certified reference material DORM-2. The percentage recoveries of total As were observed as 100.6% and 99.4% for HGAAS and GFAAS, respectively. The precision of the techniques, expressed as relative standard deviation, was observed as 1.71% and 4.18% for HGAAS and GFAAS measurements, respectively. The limits of detection for HGAAS and GFAAS were 0.025 μg/g and 0.052 μg/g, respectively. The concentrations of total arsenic in different tissues of broiler chicken were found in the range of 2.19–5.28, 2.15–5.22, 2.97–7.17 and 2.68–6.36 μg/g for leg, breast, liver and heart tissues, respectively. At a mean level of chicken consumption (60 g/person/day), people may ingest in the range of 72.0–85.1 μg arsenic/person/day from chicken alone.     

Evaluation of toxic elements in baby foods commercially available in Pakistan

In present work, the concentrations of toxic elements, aluminium (Al), cadmium (Cd), nickel (Ni) and lead (Pb) were measured in different solid baby foods (BFs), primarily to evaluate whether the intakes comply within permissible levels of these toxic elements (TEs). The BFs were evaluated for total contents of TEs, using a simple and fast ultrasound-assisted extraction (UAE) method. The accuracy of the proposed UAE method was ensured by using certified reference materials and results obtained by conventional wet acid digestion method on same CRM, at 95% confidence level. The range of the investigated TEs in different BFs were 4770–35,200, 25.6–88.3, 124–332 and 52.5–90.6 μg/kg for Al, Cd, Ni and Pb, respectively. The results indicated that BFs including rice cereals have high level of all four TEs. The daily intakes of TEs for children through BFs have also been estimated, and are well below the recommended tolerable levels.     

FIBER-MODIFIED SCALING IN HEAT TRANSFER FOULING MITIGATION

The study of heat exchanger fouling using supersaturated calcium sulphate solutions has been widely reported. In this study fouling was investigated in a larger-scale heat exchange apparatus using stainless-steel pipe, and data were obtained at different flow rates, concentrations and temperature differences. The deposits were examined using a scanning electron microscope, X-ray diffraction and conventional photography. In a novel approach, wood pulp fibers were added to the fouling solution at various concentrations to mitigate fouling. Heat transfer enhancement above the solution-alone was observed initially and the onset of fouling delayed. When fouling eventually developed the final asymptotic level was lower than the fiber-free case for the experimental conditions specified. At a fiber concentration of 0.15% heat transfer augmentation occurred for 11 days. However, at 0.25% fiber concentration, heat transfer augmentation (no fouling) was sustained over the experimental duration of 45 days. It can be concluded that the service-life cycle of a heat exchanger can be prolonged with the addition of asymmetric, flexible, natural fibers. In this work it is argued that fibers modify the onset of deposition by boundary layer scavenging, and interact with turbulent eddies to reduce the rate of mass transfer of the foulant to the heated surface. When scale forms, the crystalline structure of the scale is interrupted by the fibers, which appear to roughen the heat transfer surface initially and increase the heat transfer coefficient. However, the scale deposit continues to build up very slowly, causing the thermal resistance to eventually override the turbulence augmented heat transfer effect of the fibers.     

Dynamic optimization for gas blending in pipeline networks with gas interchangeability control

The increasing penetration of unconventional gas and liquefied natural gas poses an operational challenge on existing regional gas networks for gas quality problems. A new dynamic model for natural gas pipeline network with multiple supplies is presented with a special emphasis on gas interchangeability control. Wobbe index serves as gas interchangeability indicator and is calculated by equations derived from rigorous composition-based partial differential equations. Disjunctive formulation is applied to represent different modes of gas blending due to gas reversal, and the disjunctive model is then reformulated as a nonsmooth model with min/max and absolute value functions, which is solved by a gradient-based nonlinear program solver after smooth approximation. Moreover, a heuristic algorithm is proposed to tune the penalty parameters in order to focus on different penalty terms while keeping the model well-conditioned. The developed model and strategy are first tested with a small pipeline network model and then extended to a large model. The results show that the model can effectively manage gas interchangeability issues in pipeline networks within reasonable CPU time.