Search

Photonic delay-based reservoir computers (RC) have emerged as an attractive high-speed, low-power alternative to traditional digital hardware for AI. We demonstrate experimentally a novel hybrid RC scheme in which input data is first preprocessed through several convolutional layers, either trained or untrained, digitally to generate novel feature maps. These random feature maps are then processed through an optoelectronic implementation of delay-based RC. Using the MNIST dataset of handwritten digits, experiments of our proposed hybrid scheme achieve classification error of 1.6% using untrained convolutions, and an error of 1.1% using trained convolutions, results comparable to that of state-of-the-art machine learning algorithms. Additionally, our experimental implementation can offer a potential 10x decrease in model training time, compared to that of common digital alternatives.

The intrinsic vascular smooth muscle contraction and vasoconstriction show time-of-day variations, contributing to the blood pressure circadian rhythm, which is essential for cardiovascular health. This brief review provides an overview of our current understanding of the mechanisms underlying the time-of-day variations of vascular smooth muscle contraction. We discuss the potential contribution of the time-of-day variations of vasoconstriction to the physiological blood pressure circadian rhythm. Finally, we survey the data obtained in the type 2 diabetic db/db mouse model that demonstrate the alterations of the time-of-day variations of vasoconstriction and the non-dipping blood pressure in diabetes.

High-speed 3D printing has recently gained much interest due to its potentials in improving efficiency of fabricating complex geometry components and applications in large-scale additive manufacturing (AM). In this study, a parametric study is performed experimentally to investigate factors affecting high-speed 3D printing of continuous carbon fiber reinforced composites (CFRCs), including material deposition rate, print (nozzle traverse) speed, and nozzle tilt angle based on a novel multi-axis AM approach. The method uses thermoplastic pellets and continuous carbon fiber tows as feedstock materials. The obtained sample quality and mechanical properties are investigated with respect to deposition rate, print speed, and nozzle tilt angle. The fiber impregnation quality is examined through microstructure analysis and correlated with the process conditions and mechanical properties. Increasing deposition rate and tilt angle both improve fiber impregnation quality, enabling implementation of higher print speed and yielding improved mechanical properties. This, combined with demonstrations of printed complex geometry components, shows the great potentials of the proposed method for AM of continuous CFRCs at high speeds. The results of this study also provide further guidance on design and manufacturing of large-volume, high-strength CFRCs through 3D printing.

Aortic diameter is a standard parameter for defining disease severity of thoracic aortic aneurysms. In mouse studies, aortic diameters can be measured in situ directly, but this approach has a potential confounder of underestimation due to the absence of physiological arterial blood pressure. In the present study, we developed an in situ approach for authentic aortic measurements. Thoracic aortic aneurysms were induced by beta-aminopropionitrile (BAPN, 0.5% wt/vol) administration in 4-week-old male C57BL/6J mice. Ultrasonography was performed to examine aortic dimensions, and mice with thoracic aortic dilatations were terminated subsequently. After saline perfusion through the left ventricle, periaortic tissues were removed to expose thoracic aortas. Optimal cutting temperature (OCT) compound was injected via the left ventricle to maintain aortic patency. In situ aortic images were captured pre- and post-OCT injection. In mice with severe thoracic aortic aneurysms, smaller aortic diameters were observed prior to OCT injection compared to ultrasound measurements, while aortic diameters in situ after OCT were comparable to diameters measured using ultrasound. A telemetry system demonstrated that maximal luminal pressures during 150 ul of OCT injection were 90 mmHg. Immunostaining for CD31 revealed that endothelial cells were preserved in the intima after OCT injection. These results indicate that OCT injection does not cause aortic damages due to excess pressures. In conclusion, in situ imaging with OCT injection provides authentic aortic measurements without overt aortic damage in mice with thoracic aortic aneurysms.

Digital try-on systems for e-commerce have the potential to change people’s lives and provide notable economic benefits. However, their development is limited by practical constraints, such as accurate sizing of the body and realism of demonstrations. We enumerate three open challenges remaining for a complete and easy-to-use try-on system that recent advances in machine learning make increasingly tractable. For each, we describe the problem, introduce state-of-the-art approaches, and provide future directions.

We present a novel photonic integrated circuit (PIC) that monolithically integrates a racetrack colliding-pulse mode-locked laser with a pulse-picking electro-absorption modulator and a semiconductor optical amplifier on Indium Phosphide. We present detailed characterization of this PIC that includes optical pulse characterization, phase noise and long term stability under passive and hybrid mode-locking conditions. Allan deviation measurements made on the optical pulse train from the PIC show a fractional frequency instability of $8\times 10 ^{-11}$ at 1 second and follow a 1/ $\tau $ trend. We also demonstrate repetition rate reduction from ~10 GHz to ~500 MHz with an extinction ratio of ~14.65 dB using an on-chip pulse-picking electro-absorption modulator.

Cell differentiation agent-2 (CDA-2) was a promising hypomethylating agent approved by the Chinese FDA for the therapy of MDS in China. The active components of CDA-2 are differentiation inducers (DIs) and differentiation helper inducers (DHIs). DIs are chemicals capable of eliminating telomerase from abnormal MEs commonly found in human cancers. The major DI of CDA-2 is an organic acid without UV absorption. Without UV absorption as a guide, it was difficult to purify the DI of CDA-2 for identification. Thus, we pursued possible candidates to function as DIs in this study. Cancer MEs become abnormal due to association with telomerase. Naturally we sought telomerase inhibitors as possible candidates of DIs. Prostaglandin E2 (PGE2) attracted our attention because it was implicated to involve in wound healing, which is a major biological mission of progenitor stem cells (PSCs) and cancer stem cells (CSCs). Eradication of CSCs has been a major focus of our studies. Besides, PGE2 fits the description of the major DI of CDA-2. Induction of terminal differentiation (TD) of HL-60 cells by NBT assay was employed to evaluate the activity of chemicals as DIs. Cell growth was based on cell numbers. All-trans retinoic acid (ATRA) and 12-O-tetradecanoylphorbol-13-acetate (TPA) are two well known DIs. ATRA displayed a wide active dosage range from 0.2 to 4.5 µM with a maximum of inducing 89% NBT+ cells at 3 µM. TPA displayed a narrow active dosage range from 0.2 to 0.6 nM with a maximum of inducing 84% NBT+ cells at 0.4 nM. BIBR1532 and bodine were the two telomerase inhibitors studied. Both were found active as DIs. BIBR1532 was active in the dosage range from 30 to 75 µM with a maximum of inducing 86% at 63 µM. Bodine was active in the dosage range from 60 to 98 µM with a maximum of inducing 80% at 98 µM. PGE2 was active in the dosage range from 20 to 70 µM with a maximum of inducing 80% at 56 µM. DIs at dosages not active as DIs could function as effective DHIs to other DIs. RI0.5 of BIBR1532, boldine and PGE2 as DHIs were 2.02 µM, 3.11 µM, and 0.92 µM, respectively. DIs alone, no matter how effective, could not induce NBT+ cells to reach 100%. 95% (89% plus 6% of blank) was the highest value achieved by ATRA. Incomplete induction of TD was the reason for frequent recurrence when ATRA was used alone in the therapy of acute promyelocytic leukemia (APL). A combination of ATRA and a DHI could induce NBT+ cells to reach 100% to avoid recurrence.

Quality and efficiency are crucial indicators of any manufacturing company. Many companies are suffering from a shortage of experienced workers across the production line to perform complex assembly tasks. To reduce time and error in an assembly task, a worker-centered system consisting of multi-modal Augmented Reality (AR) instructions with the support of a deep learning network for tool detection is introduced. The integrated AR is designed to provide on-site instructions including various visual renderings with a fine-tuned Region-based Convolutional Neural Network, which is trained on a synthetic tool dataset. The dataset is generated using CAD models of tools and displayed onto a 2D scene without using real tool images. By experimenting the system to a mechanical assembly of a CNC carving machine, the result of a designed experiment shows that the system helps reduce the time and errors of the given assembly tasks by 33.2 % and 32.4 %, respectively. With the integrated system, an efficient, customizable smart AR instruction system capable of sensing, characterizing requirements, and enhancing worker’s performance has been built and demonstrated.

Silicon photonics is a rapidly maturing platform for optical communication and sensing. As systems leveraging silicon photonics have grown in size and complexity, so too has the demand for high performance silicon photonics components. In order to meet these demands, we propose a hierarchical approach to design and optimization of silicon photonics components. Our approach applies simple physical analysis to choose an effective starting geometry for a two-step gradient-based shape optimization. This optimization employs carefully chosen geometrical constraints in order to consistently produce robust, high performance devices which satisfy practical fabrication constraints of deep UV lithography. In order to demonstrate the versatility of method, we optimize a 3 dB coupler which achieves better than 0.04 dB excess loss over the O-band, a four port 3-dB coupler which achieves better than 0.41 dB excess loss and near 50:50 splitting over the O-band, and a fabrication-tolerant waveguide crossing which achieves better than 0.075 dB insertion loss over the O-band even when subject to $\pm \text{10}$ % silicon thickness variations. These results pave the way for high efficiency silicon photonic component libraries.

In a human-centered intelligent manufacturing system, every element is to assist the operator in achieving the optimal operational performance. The primary task of developing such a human-centered system is to accurately understand human behavior. In this paper, we propose a fog computing framework for assembly operation recognition, which brings computing power close to the data source in order to achieve real-time recognition. For data collection, the operator's activity is captured using visual cameras from different perspectives. For operation recognition, instead of directly building and training a deep learning model from scratch, which needs a huge amount of data, transfer learning is applied to transfer the learning abilities to our application. A worker assembly operation dataset is established, which at present contains 10 sequential operations in an assembly task of installing a desktop CNC machine. The developed transfer learning model is evaluated on this dataset and achieves a recognition accuracy of 95% in the testing experiments.

This review highlights the impact of wound onto the evolution and therapy of cancer. The objective of this review is to explore the mechanism of wound healing and its relationship to the evolution and therapy of cancer. Wound healing requires the proliferation and the terminal differentiation (TD) of progenitor stem cells (PSCs), which are the precursors of cancer stem cells (CSCs).Healing wound is not a big deal. If the functionality of chemo-surveillance is intact such as healthy people who can maintain a steady level of wound healing metabolites functioning as differentiation inducers (DIs) and differentiation helper inducers (DHIs).Wounds are always successfully healed without having to put up any effort, just to let the nature to do the healing. Medications such as suture and antibiotics are subsidiary to speed up the healing or to prevent infection. Acute wound affects the functionality of chemo-surveillance only temporarily, which is quickly recovered to return to the normal state. It is the chronic wound such as persistent infectious diseases or exposure to toxic chemicals including carcinogens for a long time that produces damaging effect on the functionality of chemo-surveillance.Chronic wound prompts the production of inflammatory cytokines to cause excessive urinary excretion of wound healing metabolites to affect wound healing. Without sufficient wound healing metabolites to terminate the proliferation of PCSs, it is very easy for PSCs to evolve into CSCs. It takes only a single hit to silence TET-1 enzyme to complete the transition, which is well within the reach of PSCs equipped with abnormally active methylation enzymes (MEs).CSCs can then progress to faster growing cancer cells by the activation of oncogenes or the inactivation of (AML). Cancer due to wound not healing properly is not unique to MDS and AML. It is rather a common phenomenon. War on cancer can be easily won if the battle is conducted following the nature’s course to heal the wound, just like the success of wound healing without having to put up any effort in healthy people. Therefore, the best strategy to win the war on cancer is to restore the functionality of chemo-surveillance by the employment of DIs and DHIs and to prevent the loss of wound healing metabolites through anti-cachexia chemicals such as phenylacetylglutamine. Then the nature will take its course to stop the proliferation of cells with abnormal MEs that include CSCs, PSCs, and all cancer cells. Destruction strategy to kill cancer cells is definitely counter indication. It creates more damages to the functionality of chemo-surveillance to stop the growth of cells with abnormal MEs. Inability of destruction strategy to put away CSCs is a deciding factor to deny the success of destruction strategy to win the war on cancer.

A high efficiency and low crosstalk SiN grating coupler was demonstrated for trapped ion qubit manipulation. By optimizing a chirped double-etch grating with inverse design, a diffraction efficiency of 98.0 % was achieved for 729 nm. Additionally, an ellipse-shaped beam resulted in a low crosstalk of -36 dB assuming 7.07 µm neighboring ion spacing.

The presence of circulating targets and antidrug antibodies can influence the ability of a bioanalytical method to measure therapeutic protein (TP) concentration relevant to exposure-response evaluations. This project surveyed biosimilar submissions for their bioanalytical methods. Survey results revealed that 97% of pharmacokinetic methods designed to measure theoretically free or partial-free TPs with respect to target indeed measured free or partial-free TPs when considering experimental testing results for target effects. Antidrug antibody effect is less often evaluated. The observed trend of measuring biologically active forms of TP is consistent with the scientific understanding that pharmacokinetics of biologically active forms is more likely to be relevant to the clinical responses and evaluation of clinically meaningful differences to contribute to biosimilarity assessments.

The rate of spontaneous emission from an optical emitter can be greatly enhanced using a metallic optical antenna at the penalty of efficiency. In this paper we propose a metal-dielectric antenna that eliminates the tradeoff between spontaneous emission enhancement and radiative efficiency by using nanoscopic dielectric structures at the antenna tips. This tradeoff occurs due to Ohmic loss and is further exacerbated by electron surface collisions. We find that our metal-dielectric antenna can enhance spontaneous emission by a factor 5 × 105 with efficiency = 70%, greatly exceeding the radiative efficiency of a purely metallic antenna with similar enhancement. Moreover, the metal-dielectric antenna design strategy is naturally amenable to short-distance optical communications applications. We go on to discuss the Purcell effect within the context of antenna enhancement. Metallic optical antennas are best analyzed with conventional antenna circuit models, but if the Purcell enhancement were to be employed, we provide the effective mode volume, Veff = (3/4π2)2 d2λ(λ/l)5, that would be needed.

The quantity and quality of food intake have been considered crucial for peoples' wellness. Only recently has it become appreciated that the timing of food intake is also critical. Nondipping blood pressure (BP) is prevalent in diabetic patients and is associated with increased cardiovascular events. However, the causes and mechanisms of nondipping BP in diabetes are not fully understood. Here, we report that food intake and BP were arrhythmic in diabetic db/db mice fed a normal chow diet ad libitum. Imposing a food intake diurnal rhythm by time-restricted feeding (TRF; food was only available for 8 h during the active phase) prevented db/db mice from developing nondipping BP and effectively restored the already disrupted BP circadian rhythm in db/db mice. Interestingly, increasing the time of food availability from 8 h to 12 h during the active dark phase in db/db mice prompted isocaloric feeding and still provided robust protection of the BP circadian rhythm in db/db mice. In contrast, neither 8-h nor 12-h TRF affected BP dipping in wild-type mice. Mechanistically, we demonstrate that TRF protects the BP circadian rhythm in db/db mice via suppressing the sympathetic activity during the light phase when they are inactive and fasting. Collectively, these data reveal a potentially pivotal role of the timing of food intake in the prevention and treatment of nondipping BP in diabetes.

Background: Cervical pregnancy (CP) is a rare form of ectopic pregnancy. Profuse and uncontrolled bleeding remains a concern when treating CP. Although many therapeutic approaches have been propose...

Obesity-related hypertension is a major public health concern. We recently demonstrated that plasma levels of the soluble form of the prorenin receptor (sPRR) were elevated in obesity-associated hypertension. Therefore, in the present study, we investigated the contribution of sPRR to blood pressure elevation in the context of obesity. High-fat fed C57BL/6 male mice were infused with vehicle or sPRR (30 µg/kg/day) via subcutaneously implanted osmotic minipump for 4 weeks. Blood pressure parameters were recorded using radiotelemetry devices. Male mice infused with sPRR exhibited higher systolic blood pressure and mean arterial pressure and lower spontaneous baroreflex sensitivity than mice infused with vehicle. To define mechanisms involved in systolic blood pressure elevation, mice were injected with an angiotensin-II type 1 receptor antagonist (losartan), a muscarinic receptor antagonist (atropine), a β-adrenergic antagonist (propranolol) and a ganglionic blocker (chlorisondamine). Losartan did not blunt sPRR-induced elevation in systolic blood pressure. Chlorisondamine treatment exacerbated the decrease in mean arterial pressure in male mice infused with sPRR. These results demonstrated that sPRR induced autonomic nervous dysfunction. Interestingly, plasma leptin levels were increased in high-fat fed C57BL/6 male mice infused with sPRR. Overall, our results indicated that sPRR increased systolic blood pressure through an impairment of the baroreflex sensitivity and an increase in the sympathetic tone potentially mediated by leptin in high-fat fed C57BL/6 male mice.

Existing therapies to improve heart function target β-adrenergic receptor (β-AR) signaling and Ca(2+) handling and often lead to adverse outcomes. This underscores an unmet need for positive inotropes that improve heart function without any adverse effects. The GTPase Ras associated with diabetes (RAD) regulates L-type Ca(2+) channel (LTCC) current (I(Ca,L)). Global RAD-knockout mice (gRAD(−/−)) have elevated Ca(2+) handling and increased cardiac hypertrophy, but RAD is expressed also in noncardiac tissues, suggesting the possibility that pathological remodeling is due also to noncardiac effects. Here, we engineered a myocardial-restricted inducible RAD-knockout mouse (RAD(Δ/Δ)). Using an array of methods and techniques, including single-cell electrophysiological and calcium transient recordings, echocardiography, and radiotelemetry monitoring, we found that RAD deficiency results in a sustained increase of inotropy without structural or functional remodeling of the heart. I(Ca,L) was significantly increased, with RAD loss conferring a β-AR–modulated phenotype on basal I(Ca,L). Cardiomyocytes from RAD(Δ/Δ) hearts exhibited enhanced cytosolic Ca(2+) handling, increased contractile function, elevated sarcoplasmic/endoplasmic reticulum calcium ATPase 2 (SERCA2a) expression, and faster lusitropy. These results argue that myocardial RAD ablation promotes a beneficial elevation in Ca(2+) dynamics, which would obviate a need for increased β-AR signaling to improve cardiac function.

The new and rapid advancement in the complexity of biologics drug discovery has been driven by a deeper understanding of biological systems combined with innovative new therapeutic modalities, paving the way to breakthrough therapies for previously intractable diseases. These exciting times in biomedical innovation require the development of novel technologies to facilitate the sophisticated, multifaceted, high‐paced workflows necessary to support modern large molecule drug discovery. A high‐level aspiration is a true integration of “lab‐on‐a‐chip” methods that vastly miniaturize cellulmical experiments could transform the speed, cost, and success of multiple workstreams in biologics development. Several microscale bioprocess technologies have been established that incrementally address these needs, yet each is inflexibly designed for a very specific process thus limiting an integrated holistic application. A more fully integrated nanoscale approach that incorporates manipulation, culture, analytics, and traceable digital record keeping of thousands of single cells in a relevant nanoenvironment would be a transformative technology capable of keeping pace with today's rapid and complex drug discovery demands. The recent advent of optical manipulation of cells using light‐induced electrokinetics with micro‐ and nanoscale cell culture is poised to revolutionize both fundamental and applied biological research. In this review, we summarize the current state of the art for optical manipulation techniques and discuss emerging biological applications of this technology. In particular, we focus on promising prospects for drug discovery workflows, including antibody discovery, bioassay development, antibody engineering, and cell line development, which are enabled by the automation and industrialization of an integrated optoelectronic single‐cell manipulation and culture platform. Continued development of such platforms will be well positioned to overcome many of the challenges currently associated with fragmented, low‐throughput bioprocess workflows in biopharma and life science research.

Purpose: In this paper, we describe a method for recovering the tissue properties directly from medical images and study the correlation of tissue (i.e., prostate) elasticity with the aggressiveness of prostate cancer using medical image analysis. Methods: We present a novel method that uses geometric and physical constraints to deduce the relative tissue elasticity parameters. Although elasticity reconstruction, or elastograph, can be used to estimate tissue elasticity, it is less suited for in-vivo measurements or deeply seated organs like prostate. We develop a method to estimate tissue elasticity values based on pairs of images, using a finite-element-based biomechanical model derived from an initial set of images, local displacements, and an optimization-based framework. Results: We demonstrate the feasibility of a statistically based classifier that automatically provides a clinical T-stage and Gleason score based on the elasticity values reconstructed from computed tomography images. Conclusion: We study the relative elasticity parameters by performing cancer grading/staging prediction and achieve up to 85% accuracy for cancer staging prediction and up to 77% accuracy for cancer grading prediction using a feature set, which includes recovered relative elasticity parameters and patient age information.

An important aspect of dynamic urban coverage is how building collision avoidance is incorporated into the overall coverage mission. We consider a multi-agent urban dynamic coverage problem in which a team of flying agents uses downward facing cameras to observe the street-level environment outside of buildings. Cameras are assumed to be ineffective above a maximum altitude (lower than building height), such that agents must move around or over buildings to complete their mission. The main objective of this paper is to compare three different building avoidance strategies that are compatible with dynamic ergodic methods. To provide context for these results, we also compare our results to three other common coverage methods including: boustrophedon coverage (lawn-mower sweep), Voronoi region based coverage, and a naive grid method. All algorithms are evaluated in simulation with respect to four performance metrics (percent coverage, revisit count, revisit time, and the integral of area viewed over time), across team sizes ranging from 1 to 25 agents, and in five types of urban environments of varying density and height. We find that the relative performance of algorithms changes based on the ratio of team size to search area, as well the height and density characteristics of the urban environment.

Autonomous systems often rely on neural networks to achieve high performance on planning and control problems. Unfortunately, neural networks suffer severely when input images become degraded in ways that are not reflected in the training data. This is particularly problematic for robotic systems like autonomous vehicles (AV) for which reliability is paramount. In this work, we consider robust optimization methods for hardening control systems against image corruptions and other unexpected domain shifts. Recent work on robust optimization for neural nets has been focused largely on combating adversarial attacks. In this work, we borrow ideas from the adversarial training and data augmentation literature to enhance robustness to image corruptions and domain shifts. To this end, we train networks while augmenting image data with a battery of image degradations. Unlike traditional augmentation methods, we choose the parameters for each degradation adversarially so as to maximize system performance. By formulating image degradations in a way that is differentiable with respect to degradation parameters, we enable the use of efficient optimization methods (PGD) for choosing worst-case augmentation parameters. We demonstrate the efficacy of this method on the learning to steer task for AVs. By adversarially training against image corruptions, we produce networks that are highly robust to image corruptions. We show that the proposed differentiable augmentation schemes result in higher levels of robustness and accuracy for a range of settings as compared to baseline and state-of-the-art augmentation methods.

We report on a co-planar light-actuated digital microfluidics device that features a metal mesh grid integrated on the device surface to allow droplets to be exposed from above. We discuss a theoretical circuit model for our co-planar optoelectrowetting (OEW) design that allows for the optimization of droplet actuation while maintaining reliable droplet movement. Basic droplet manipulations such as merging and parallel actuation of droplets are achieved at speeds of up to 4.5 cm / s. The co-planar OEW device design benefits from having an open top design that allows for a wider range of system integration configurations than previous generations of OEW devices. A droplet-on-demand dispensing system from above is integrated with the co-planar OEW device to demonstrate the versatility of this optofluidic platform. The ability to inject, collect, and position individual droplets to form large-scale droplet arrays of up to 20 × 20 is achieved.

Background and Objectives: Hyperinsulinemia plays a key role in development of cardiovascular impairment in patients with Metabolic syndrome. The aim of this study was to evaluate the influence of hyperinsulinemia on long-term Clinical outcomes of percutaneous coronary intervention (PCI) in patients with acute myocardial syndrome. Methods: Between March 2016 and January 2019, patients of ACS without diabetes mellitus and received primary PCI were enrolled. 368 patients were divided into low insulin group (n=157), medium insulin group (n=154) and high insulin group (n=157) according to tertiles of fasting insulin (FINS) level. The primary endpoint was major adverse cardiac events (MACE; all-cause death, non-fatal myocardial infarction, target vessel revascularization (TVR)) at 24 months. Second endpoint was angina-hospitalization. Results: High insulin group had an unfavorable prognosis, with higher rate of MACE (34.39%) compared with low insulin group (22.29%) and medium insulin group (23.37%) at 24 months (PConclusion: Hyperinsulinemia might be a valid predictor of clinical outcomes in ACS patients undergoing PCI.

We report on a 32 × 32 silicon photonic micro-electro-mechanical-system (MEMS) switch with gap-adjustable directional couplers. The switch is fabricated on 200-mm silicon-on-insulator wafers in a commercial complementary metal-oxide-semiconductor (CMOS) foundry. The fabricated device has a maximum on-chip loss of 7.7 dB and an extinction ratio of 50.8 dB. The switching voltage is 9.45 V and the 20-dB bandwidth is 28.7 nm. Our work shows a promising path for mass production of silicon photonic MEMS switches in commercial CMOS foundries.

Reservoir computing has emerged as a lightweight, high-speed machine learning paradigm. We introduce a new optoelectronic reservoir computer for image recognition, in which input data is first pre-processed offline using two convolutional neural network layers with randomly initialized weights, generating a series of random feature maps. These random feature maps are then multiplied by a random mask matrix to generate input nodes, which are then passed to the reservoir computer. Using the MNIST dataset in simulation, we achieve performance in line with state-of-the-art convolutional neural networks (1% error), while potentially offering order-of-magnitude improvement in training speeds.

In this paper, we present the development of two ultra-wideband (UWB) inter-digital cavity filters manufactured with an additive technique. We employed a selective laser melting process with stainless steel via a "flipping and fixturing" approach to achieve dimensional accuracy and frequency selectivity (no post-fabrication tuning) comparable to that of conventional micromachining. Both filters provide 2 GHz of instantaneous bandwidth. Their center operating frequencies are 3.75 GHz and 6 GHz, respectively. This paper presents design and fabrication details, a discussion of the measured characteristics of both filters, and a performance comparison with respect to subtractively manufactured parts.

We report on novel dual-layer grating couplers that support polarization diversity in silicon photonics. High coupling efficiency (-0.22 dB or 95.1%) and low polarization dependence (0.1 dB) have been realized through inverse design.

The state-of-the-art of fully-supervised methods for temporal action localization from untrimmed videos has achieved impressive results. Yet, it remains unsatisfactory for the weakly-supervised temporal action localization, where only video-level action labels are given without the timestamp annotation on when the actions occur. The main reason comes from that, the weakly-supervised networks only focus on the highly discriminative frames, but there are some ambiguous frames in both background and action classes. The ambiguous frames in background class are very similar to the real actions, which may be treated as target actions and result in false positives. On the other hand, the ambiguous frames in action class which possibly contain action instances, are prone to be false negatives by the weakly-supervised networks and result in a coarse localization. To solve these problems, we introduce a novel weakly-supervised Action Completeness Modeling with Background Aware Networks (ACM-BANets). Our Background Aware Network (BANet) contains a weight-sharing two-branch architecture, with an action guided Background aware Temporal Attention Module (B-TAM) and an asymmetrical training strategy, to suppress both highly discriminative and ambiguous background frames to remove the false positives. Our action completeness modeling contains multiple BANets, and the BANets are forced to discover different but complementary action instances to completely localize the action instances in both highly discriminative and ambiguous action frames. In the i-th iteration, the i-th BANet discovers the discriminative features, which are then erased from the feature map. The partially-erased feature map is fed into the (i+1)-th BANet of the next iteration to force this BANet to discover discriminative features different from the i-th BANet. Evaluated on two challenging untrimmed video datasets, THUMOS14 and ActivityNet1.3, our approach outperforms all the current weakly-supervised methods for temporal action localization.

We present an evanescent coupler from a fiber-coupled glass interposer to silicon photonics with integrated polarization splitter. The insertion loss in O-band is better than −0.5 dB with ±20 nm silicon thickness tolerance and 2µm epoxy thickness tolerance.

Obesity is a major risk factor for hypertension. Although the renin-angiotensin system (RAS) contributes to the sex difference of blood pressure (BP) control, whether the prorenin receptor (PRR) and its soluble form (sPRR) play a role in the sexual dimorphism of BP is not clear. We previously demonstrated that, in high fat (HF)-fed male C57BL/6 mice, the infusion of mouse recombinant sPRR increased systolic blood pressure (SBP) by the sympatho-excitatory effects of leptin on BP. Therefore, in the present study, we aim to address whether mouse recombinant sPRR influences the BP in HF-fed female mice. To test this hypothesis, C57BL/6 female mice were fed a HF diet for 32 weeks and were implanted with radiotelemetry transmitter. After 24 weeks of high fat feeding, female mice (5-6/group) were implanted with osmotic pumps and infused with either saline (veh) or sPRR for 3 to 4 weeks. In contrast to male mice, the infusion of sPRR (30 μg/kg/day) did not change significantly the SBP (24hSBP; veh: 135 ± 7; sPRR: 134 ± 4 mmHg; P>0.05) and the baroreflex sensitivity. The decrease in BP mediated by chlorisondamine treatment was not significantly different between female mice infused with vehicle or sPRR (ΔSBP; veh: -11 ± 9; sPRR: -16 ± 7 mmHg; P>0.05). In addition, sPRR infusion did not affect the bradycardic or tachycardic responses after propranolol or atropine treatment respectively. Moreover, the decrease of SBP induced by losartan was similar in mice infused with vehicle or sPRR (ΔSBP; veh: -9 ± 4; sPRR: -10 ± 3 mmHg; P>0.05). Similar results were obtained using higher dose of sPRR (60 μg/kg/day). In female mice, sPRR infusion did not increase significantly the body weight (veh: 32.7 ± 2.5 g; sPRR: 35.8 ± 4.2 g; P>0.05), the white adipose tissue weight (WAT; veh: 3.0 ± 0.7g; sPRR: 3.7 ± 1.6 g; P>0.05) or circulating leptin levels (veh: 12.0 ± 7.6 ng/ml; sPRR: 15.5 ± 9.5 ng/ml; P>0.05). In conclusion, in contrast to male, female mice are protected from sPRR-induced increase in BP. sPRR did not increase circulating leptin suggesting that sPRR-induced leptin increase is regulated in a sex-dependent manner. Other pathways could participate to the protection against sPRR-induced increase in BP in female mice such as the vasodilator arm of the RAS and/or the hormonal status of the female mice.

Efficient high speed nanoscale optical sources are required for low power next generation data communication. In this report, we experimentally demonstrated a waveguide coupling ratio of 85.9% for the cavity-backed slot antenna LED to a single mode InP waveguide.

Realistic locomotion in a virtual environment (VE) can help maximize immersion and decrease simulator sickness. Redirected walking (RDW) allows a user to physically walk in VR by rotating the VE as a function of head rotation such that they walk in an arc that fits in the tracking area. However, this requires significant user rotation, often requiring a “distractor” to cause such rotation in commercial tracking spaces. Previous implementations suddenly spawned a distractor (e.g. butterfly) when the user walks near the safe boundary, with limitations like the user causing distraction accidentally by looking around, the distractor not being acknowledged, or getting “stuck” in a corner. We explore a persistent, robot distractor tethered to the user that provides two-way haptic feedback and natural motion constraints. We design a dynamic robot AI which adapts to randomness in the user’s behavior, as well as trajectory changes caused by tugging on its leash. The robot tries to imperceptibly keep the user safe by replicating a real dog’s behaviors, such as barking or sniffing something. We hypothesize that the naturalness of the dog behavior, its responses to the user, and the haptic tethering will work together to allow the user to explore the entire city, ideally without noticing that the dog is a robot.

The quantity and quality of food intake have been considered crucial for peoples’ wellness. Only recently has it become appreciated that the timing of food intake, independent of total calorie consumption, is also critical. The involvement and contribution of the timing of food intake in nondipping blood pressure (BP), the most common disruption of the BP circadian rhythm in diabetes, however, remains uncertain. Here, we demonstrate that the loss of diurnal rhythm in food intake coincided with nondipping BP in type 2 diabetic db/db mice, and imposing a food intake diurnal rhythm by active time-restricted feeding (ATRF) remarkably prevented db/db mice from the development of nondipping BP. Moreover, ATRF, independent of calorie restriction, also effectively restored the already disrupted BP circadian rhythm in db/db mice. Mechanistically, ATRF reduced the sympathetic vascular tone and BP during the light phase via fasting, thus protecting the BP circadian rhythm in db/db mice. Moreover, we identified BMAL1, an obligatory core clock gene, as a potentially key molecule that links ATRF, sympathetic vascular tone, and BP circadian rhythm. Collectively, these data reveal an important but previously unrecognized role of the timing of food intake in the onset, prevention, and treatment of nondipping BP in diabetes.

PURPOSE Nonadherence to aromatase inhibitors (AIs) for breast cancer is common and increases the risk of recurrence. Text messaging increases adherence to medications for chronic conditions. METHODS We conducted a randomized clinical trial of text messaging (TM) versus no text messaging (No-TM) at 40 sites in the United States. Eligible patients were postmenopausal women with early-stage breast cancer taking an AI for > 30 days with a planned duration of ≥ 36 months. Test messages were sent twice a week over 36 months. Content themes focused on overcoming barriers to medication adherence and included cues to action, statements related to medication efficacy, and reinforcements of the recommendation to take AIs. Both groups were assessed every 3 months. The primary outcome was time to adherence failure (AF), where AF was defined as urine AI metabolite assay results satisfying one of the following: < 10 ng/mL, undetectable, or no submitted specimen. A stratified log-rank test was conducted. Multiple sensitivity analyses were performed. RESULTS In total, 724 patients were registered between May 2012 and September 2013, among whom,702 patients (348 in the text-messaging arm and 354 in the no–text-messaging arm) were eligible at baseline. Observed adherence at 36 months was 55.5% for TM and 55.4% for No-TM. The primary analysis showed no difference in time to AF by arm (3-year AF: 81.9% TM v 85.6% No-TM; HR, 0.89 [95% CI, 0.76 to 1.05]; P = .18). Multiple time to AF sensitivity analyses showed similar nonsignificant results. Three-year self-reported time to AF (10.4% v 10.3%; HR, 1.16 [95% CI, 0.69 to 1.98]; P = .57) and site-reported time to AF (21.9% v 18.9%; HR, 1.31 [95% CI, 0.86 to 2.01]; P = .21) also did not differ by arm. CONCLUSION To our knowledge, this was the first large, long-term, randomized trial of an intervention directed at improving AI adherence. We found high rates of AI AF. Twice-weekly text reminders did not improve adherence to AIs. Improving long-term adherence will likely require personalized and sustained behavioral interventions.

Existing state-of-the-art object tracking can run into challenges when objects collide, occlude, or come close to one another. These visually based trackers may also fail to differentiate between objects with the same appearance but different materials. Existing methods may stop tracking or incorrectly start tracking another object. These failures are uneasy for trackers to recover from since they often use results from previous frames. By using audio of the impact sounds from object collisions, rolling, etc., our audio-visual object tracking (AVOT) neural network can reduce tracking error and drift. We train AVOT end to end and use audio-visual inputs over all frames. Our audio-based technique may be used in conjunction with other neural networks to augment visually based object detection and tracking methods. We evaluate its runtime frames-per-second (FPS) performance and intersection over union (IoU) performance against OpenCV object tracking implementations and a deep learning method. Our experiments, using the synthetic Sound-20K audio-visual dataset, demonstrate that AVOT outperforms single-modality deep learning methods, when there is audio from object collisions. A proposed scheduler network to switch between AVOT and other methods based on audio onset maximizes accuracy and performance over all frames in multimodal object tracking.

Transition metal dichalcogenide (TMDC) monolayers are promising materials for nanoscale light emitters. By coupling TMDCs to optical antennas, the spontaneous emission rate can be greatly enhanced, yielding faster device modulation. However, achieving high rate enhancement is difficult with electrically-injected light emitters compared to optical injection. We demonstrate two device designs to overcome these challenges: 1) a light-emitting capacitor (LEC) coupled to a slot antenna array, and 2) a light-emitting diode (LED) coupled to a nanosquare antenna array. The LEC shows high polarization ratios >30x, while the LED shows >10x enhanced electroluminescence for the antenna coupled device relative to control devices.

Biologic-device combination products using prefilled syringes (PFSs) and autoinjectors (AIs) are popular for biological products administered subcutaneously. Pharmacokinetic (PK) comparability studies commonly provide the scientific data to support introduction of AI presentations via bridging with PFS. A survey of biological products approved by FDA's Center for Drug Evaluation and Research identified 17 biologics license applications (BLAs) with both PFS and AI presentations for subcutaneous (SC) administration, including 16 approved on February 1, 2018, and one with AI presentation under review. A systematic review on the device parameters and the PK comparability studies bridging the two presentations was conducted. Subsequently, whether device parameters or the PK study design may have influenced the PK comparability study results was evaluated. The reported device parameters for AI and PFS are generally consistent across BLAs, whereas the approach to assess PK comparability varied, including the study design. Most PK comparability studies met bioequivalence (BE) criteria. Upon inspection of the studies that did not meet BE criteria, injection depth of AI and the injection site for either AI or PFS were identified as potential influencing factors to the outcome of PK comparability study. This study represents an initial attempt to identify the potential influencing factors on device bridging, including the characteristics of the device and the clinical pharmacology study. These findings may inform the combination product development strategy, specifically design considerations for device and PK comparability studies.

The generation of realistic apparel model has become increasingly popular as a result of the rapid pace of change in fashion trends and the growing need for garment models in various applications such as virtual try-on. For such application requirements, it is important to have a general cloth model that can represent a diverse set of garments. Previous studies often make certain assumptions about the garment, such as the topology or suited body shape. We propose a unified method using the generative network. Our model is applicable to different garment topologies with different sewing patterns and fabric materials. We also develop a novel image representation of garment models, and a reliable mapping algorithm between the general garment model and the image representation that can regularize the data representation of the cloth. Using this special intermediate image representation, the generated garment model can be easily retargeted to another body, enabling garment customization. In addition, a large garment appearance dataset is provided for use in garment reconstruction, garment capturing, and other applications. We demonstrate that our generative model has high reconstruction accuracy and can provide rich variations of virtual garments.

We present a two-dimensional random-access optical beam steering system composed of a 20x20 focal plane switch array integrated on a silicon photonics chip with microelectromechanical-system (MEMS) optical switches. 32°x32°field-of-view is demonstrated.

We demonstrate the first-ever silicon photonics enabled hyper-wideband RF spread-spectrum link. A hybrid III/V-silicon photonic mode-locked laser encoded with a four-channel silicon photonic phase encoder is used to generate optical carriers that coherently demodulate a received RF signal spanning an instantaneous bandwidth greater than 85% of the center frequency (12/14 GHz). This allows low-speed (

Most recent garment capturing techniques rely on acquiring multiple views of clothing, which may not always be readily available, especially in the case of pre-existing photographs from the web. As an alternative, we propose a method that is able to compute a 3D model of a human body and its outfit from a single photograph with little human interaction. Our algorithm is not only able to capture the global shape and overall geometry of the clothing, it can also extract the physical properties (i.e., material parameters needed for simulation) of cloth. Unlike previous methods using full 3D information (i.e., depth, multi-view images, or sampled 3D geometry), our approach achieves garment recovery from a single-view image by using physical, statistical, and geometric priors and a combination of parameter estimation, semantic parsing, shape/pose recovery, and physics-based cloth simulation. We demonstrate the effectiveness of our algorithm by re-purposing the reconstructed garments for virtual try-on and garment transfer applications and for cloth animation on digital characters.

Diabetic patients have an increased prevalence of blood pressure (BP) circadian rhythm disruption, which is associated with increased risk of target organ damage and detrimental cardiovascular events. Limited information is available regarding the role of clock genes in the disruption of the BP circadian rhythm in diabetes due to the lack of a diabetic animal model that allows real-time monitoring of clock gene oscillation. Here, we generated a novel diabetic db/db-mPer2(Luc) mouse model by crossing the type 2 diabetic db/db mice with the mPer2(Luc) knock-in mice. The daily rhythms of BP, heart rate, locomotor activity, and food and water intake were acquired by radiotelemetry or metabolic chambers. The daily oscillation of mPer2 bioluminescence was recorded by LumiCycle in real-time in tissue explants and by IVIS system in vivo. Our results showed that the db/db-mPer2(Luc) mice were obese, diabetic and glucose intolerant. The db/db-mPer2(Luc) mice displayed a compromised BP daily rhythm, which was associated with the disruption of the daily rhythms in baroreflex sensitivity, locomotor activity, and metabolism, but not heart rate or food and water intake. The phase of the mPer2 daily oscillation was advanced to different extents in the explanted peripheral tissues from the db/db-mPer2(Luc) mice relative to that in the control mice. In contrast, no phase shift was detected in the mPer2 daily oscillation in the explanted suprachiasmatic nucleus (SCN). Moreover, the advanced phase shift of the mPer2 daily oscillation was also detected in the liver, kidney and submandibular gland in vivo in the db/db-mPer2(Luc) mice. In conclusion, the diabetic db/db-mPer2(Luc) mouse is a novel animal model that allows real-time monitoring of mPer2 circadian rhythms ex vivo and in vivo. The results from db/db-mPer2(Luc) mice suggest that the desynchrony of mPer2 daily oscillation in the peripheral tissues contributes to the loss of BP daily oscillation in diabetes.

A novel injection locking architecture is demonstrated to simultaneously improve the axial mode linewidth and stabilize the repetition rate of a monolithically integrated mode-locked laser. First, linewidth reduction is demonstrated via multitone injection locking. Then, the passive mode-locked laser is used as a source to make a coupled optoelectronic oscillator, and its RF spectral characteristics are investigated for different optical delays. A novel combination of these two techniques is established by creating a coupled optoelectronic oscillator via multitone injection locking, using an intracavity element as photodetector to generate the feedback signal to a Mach-Zehnder modulator. Using this combination of methods, the axial mode linewidth is reduced by over 6000× to $\sim$ 100 kHz and the RF linewidth is reduced by a factor of 70 at $-$ 30 dBc, resulting in a 3 dB RF linewidth of 400 Hz. The improvement in RF phase noise at 200 kHz offset is $>$ 40 dB. Finally, the system is referenced to a RF synthesizer using a variable voltage phase-shifter and a PID controller. Allan deviation measurements show a resolution limited stability in the repetition rate of $10^{-10}$ at 1 second following a $1/\tau$ trend.