Search

Four prenylated indole alkaloids (1−4) were targeted isolated from the mangrove rhizosphere soil-derived fungus Penicillium janthinellum HK1-6 by using molecular networking strategies. Among them, the planar structure and relative configuration of notoamide X (1) were elucidated by detailed analysis of the spectroscopic data especially the NOESY spectrum for the first time and its absolute configuration was determined by ECD spectrum. Furthermore, curated molecular networks of MS/MS data were generated with GNPS which allowed highlighting six prenylated indole alkaloids (5, 6, 8, 9, 11, 12) that had not previously been identified in this fungus and two (7, 10) that had never been observed in any fungus. The MS/MS fragmentation pathway of these prenylated indole alkaloids was summarized.

Machine learning (ML) is getting more pervasive. Wide adoption of ML in healthcare, facial recognition, and blockchain involves private and sensitive data. One of the most promising candidates for inference on encrypted data, termed Fully Homomorphic Encryp-tion (FHE), preserves the privacy of both data and the ML model. However, it slows down plaintext inference by six magnitudes, with a root cause of replacing non-polynomial operators with latency-prohibitive 27-degree Polynomial Approximated Function (PAF). While prior research has investigated low-degree PAFs, naive stochastic gradient descent (SGD) training fails to converge on PAFs with degrees higher than 5, leading to limited accuracy compared to the state-of-the-art 27-degree PAF. Therefore, we propose four training techniques to enable convergence in the post-approximation model using PAFs with an arbitrary degree, including (1) Dynamic Scaling (DS) and Static Scaling (SS) to enable minimal approximation error during approximation, (2) Coefficient Tuning (CT) to obtain a good initial coefficient value for each PAF, (3) Progressive Approximation (PA) to simply the two-variable regression optimization problem into single-variable for fast 1 and easy convergence, and (4) Alternate Training (AT) to retraining the post-replacement PAFs and other linear layers in a decoupled divide-and-conquer manner. A combination of DS/SS, CT, PA, and AT enables the exploration of accuracy-latency space for FHE-domain ReLU replacement. Leveraging the proposed techniques, we propose a systematic approach (PAF-FHE) to enable low-degree PAF to demonstrate the same accuracy as SotA high-degree PAFs. We evaluated PAFs with various degrees on different models and variant datasets, and PAF-FHE consistently enables low-degree PAF to achieve higher accuracy than SotA PAFs. Specifically, for ResNet-18 under the ImageNet-1k dataset, our spotted optimal 12-degree PAF reduces 56% latency compared to the SotA 27-degree PAF with the same post-replacement accuracy (69.4%). While as for VGG-19 under the CiFar-10 dataset, optimal 12-degree PAF achieves even 0.84% higher accuracy with 72% latency saving. Our code is open-sourced at: https://github.com/TorchFHE/PAF-FHE

Carbon deposition is the main cause for the catalyst deactivation of methane dry reforming, and researchers are committed to exploring effective catalyst systems with zero carbon deposition in order to achieve a practically long life-time. In this work, we propose an equilibrium theory with matched rates of CH4 dissociation and CO2 activation to establish a balance between carbon deposition and carbon elimination, and construct highly dispersed Ni-Ir/MgAl2O4 alloy catalysts accordingly, where Ni activated CH4, MgAl2O4 adsorbed CO2 to form surface carbonates, and Ir effectively utilized the carbonates to eliminate carbon species generated by CH4 dissociation. Theoretical assessment further unveiled that the preferred CO2 activation on Ir over Ni is derived from its stronger oxophilicity. With an optimal Ni/Ir atomic ratio of 1/2, high activity and long-period stability (600 h) with zero carbon deposition were obtained concurrently for dry reforming of methane at industrially-relevant temperature (650 ℃).

Conversion of naturally occurring sugars, the most abundant biomass resources on Earth, to fuels and chemicals provides a sustainable and carbon-neutral alternative to the current fossil resource–based processes. Tungsten-based catalysts (e.g., WO 3 ) are efficient for selectively cleaving C-C bonds of sugars to C 2,3 oxygenate intermediates (e.g., glycolaldehyde) that can serve as platform molecules with high viability and versatility in the synthesis of various chemicals. Such C-C bond cleavage follows a mechanism distinct from the classical retro-aldol condensation. Kinetic, isotope 13 C-labeling, and spectroscopic studies and theoretical calculations, reveal that the reaction proceeds via a surface tridentate complex as the critical intermediate on WO 3 , formed by chelating both α- and β-hydroxyls of sugars, together with the carbonyl group, with two adjacent tungsten atoms (W-O-W) contributing to the β-C-C bond cleavage. This mechanism provides insights into sugar chemistry and enables the rational design of catalytic sites and reaction pathways toward the efficient utilization of sugar-based feedstocks.

Significant growth is seen in artificial intelligence (AI), in particular deep learning (DL), which has made remarkable progress in various areas such as computer vision, natural language processing, health care, autonomous driving, and surveillance. To accomplish this, AI technologies have broadened from a centralized fashion to mobile or distributed fashion, opening a new era called edge AI, with dramatic advancements that are substantially changing everyday technology, social behavior, and lifestyles. Edge AI couples intelligence and analysis to a broad collection of connected devices and systems for data collection, caching, and processing. It enables a wide variety of new promising applications where data collection and analysis are combined. Billions of mobile users are exploiting various smartphone applications such as translation services, digital assistants, and health monitoring services.

In this paper, we propose a data-model-hardware tri-design framework for high-throughput, low-cost, and high-accuracy multi-object tracking (MOT) on High-Definition (HD) video stream. First, to enable ultra-light video intelligence, we propose temporal frame-filtering and spatial saliency-focusing approaches to reduce the complexity of massive video data. Second, we exploit structure-aware weight sparsity to design a hardware-friendly model compression method. Third, assisted with data and model complexity reduction, we propose a sparsity-aware, scalable, and low-power accelerator design, aiming to deliver real-time performance with high energy efficiency. Different from existing works, we make a solid step towards the synergized software/hardware co-optimization for realistic MOT model implementation. Compared to the state-of-the-art MOT baseline, our tri-design approach can achieve 12.5x latency reduction, 20.9x effective frame rate improvement, 5.83x lower power, and 9.78x better energy efficiency, without much accuracy drop., Accepted to ASP-DAC'23

Research background and Objectives: Age is an independent risk factor for cardiovascular disease (CVD), but CVD risk factors are preventable, and lack of awareness of its risk factors is a contributing factor to CVDs. Middle-aged people may be more likely to engage in unhealthy lifestyle behaviours which can increase the risk of CVD. Health self-assessment is crucial for early detection and management of health issues and early lifestyle intervention for better personalised health management. This study aims to determine the self-assessment of INTERHEART risk classification among the middle-aged community in Malaysia. Method: Local community members aged 40–60 years and who are currently residing in Malaysia were recruited via non-randomised sampling. Sociodemographic characteristics and dietary pattern related to salt, fibre, fat (deep fried/snacks), poultry/meat intakes, and other cardiovascular risk factors (waist-hip ratio, medical history related to diabetes/hypertension, history/exposure of tobacco use, psychosocial status, and level of physical activity) were assessed; INTERHEART risk scores were then computed and stratified into low, medium and high risks. Results: Approximately 45% (n = 273/602) of middle-aged respondents in Malaysia are at moderate-to-high risk of cardiovascular events, with men being more likely to develop CVD compared to women. The results of the survey indicated that poultry/meat intake (61%), physical inactivity (59%), and second-hand smoke (SHS) exposure (54%) are the most prevalent risk factors among the respondents. One-third of the respondents consumed excessive salty food and deep fried foods/snacks/fast food, and only one-third of them consumed vegetables/fruits at a recommended level. It is worrying that about a quarter of the respondents felt several periodical/permanent stresses and even felt sad/blue/depressed for two weeks or more in a row. Males, labour workers, and those with lower educational levels are more likely to develop CVD events. Conclusions: This study found that 45% of the middle-aged respondents were having moderate-to-high risk for cardiovascular events with multiple risk factors related to unhealthy lifestyle habits and environmental factors. In addition to non-modifiable factors such as gender and age, sociodemographic factors, i.e., educational level and occupation, are equally important factors to determine CVD risk. Overall, the findings of this study emphasize the clinical relevance of assessing multiple factors in the determination of CVD risks for early prevention and management of cardiovascular diseases.

Introduction: The utilization of fossil fuels in coal-fired power plants (CFPPs) for electricity generation could lead to the release of additional natural and artificial radioisotopes into the surrounding environment and consequently, increase the gamma dose rate at these areas. Methods: In this study, the geological variation of gamma dose rate in the vicinity of Duyen Hai CFPP complex and Vinh tan CFPP complex were evaluated. Results: The obtained gamma dose rates were in the range from 0.07 to 0.22 μSv h-1 with the average value of 0.13 μSv h-1 for Duyen Hai area and in the range from 0.14 to 0.34 μSv h-1 with the average value of 0.21 μSv h-1. The results indicated that the gamma dose rate in Vinh tan area is higher than the corresponding value of Duyen Hai area. Both natural aspects, such as the geological differences as well as geomorphology and weathering process, and anthropogenic aspects, such as the difference in operational scale, coal consumption and ash discharge between Duyen Hai and Vinh Tan CFPP complex, might significantly contribute to the variation of gamma dose rate in the investigated locations. Conclusion: Overall, the calculated outdoor annual effective doses of two studied areas showed that the radiological impacts of both Duyen Hai and Vinh Tan CFPP complex to the environment and public health are negligible.

This study experimentally investigates the effect of the opening configuration on the buckling stability and bearing performance of a structural beam web used in a commercial aircraft made of composite materials. The buckling and strength analyses on three opening configurations (circular, oblong, and rhombic) were carried out using test samples with identical web surface size. It is found that the rhombic opening has the minimum effect on the buckling stability and strength of the structure. To compensate for the effect of the opening, two reinforcement methods, using reinforcement rib and thickening the sample, were also investigated in this study. It is concluded that thickening the sample can more effectively improve the buckling stability and strength performance of beam web structure and hence has relatively higher structural reinforcement efficiency.

The complex nature of real-world problems calls for heterogeneity in both machine learning (ML) models and hardware systems. The heterogeneity in ML models comes from multi-sensor perceiving and multi-task learning, i.e., multi-modality multi-task (MMMT), resulting in diverse deep neural network (DNN) layers and computation patterns. The heterogeneity in systems comes from diverse processing components, as it becomes the prevailing method to integrate multiple dedicated accelerators into one system. Therefore, a new problem emerges: heterogeneous model to heterogeneous system mapping (H2H). While previous mapping algorithms mostly focus on efficient computations, in this work, we argue that it is indispensable to consider computation and communication simultaneously for better system efficiency. We propose a novel H2H mapping algorithm with both computation and communication awareness; by slightly trading computation for communication, the system overall latency and energy consumption can be largely reduced. The superior performance of our work is evaluated based on MAESTRO modeling, demonstrating 15%-74% latency reduction and 23%-64% energy reduction compared with existing computation-prioritized mapping algorithms., 6 pages

Graph neural networks (GNNs) have recently exploded in popularity thanks to their broad applicability to graph-related problems such as quantum chemistry, drug discovery, and high energy physics. However, meeting demand for novel GNN models and fast inference simultaneously is challenging due to the gap between developing efficient accelerators and the rapid creation of new GNN models. Prior art focuses on accelerating specific classes of GNNs, such as Graph Convolutional Networks (GCN), but lacks generality to support a wide range of existing or new GNN models. Furthermore, most works rely on graph pre-processing to exploit data locality, making them unsuitable for real-time applications. To address these limitations, in this work, we propose a generic dataflow architecture for GNN acceleration, named FlowGNN, which is generalizable to the majority of message-passing GNNs. The contributions are three-fold. First, we propose a novel and scalable dataflow architecture, which generally supports a wide range of GNN models with message-passing mechanism. The architecture features a configurable dataflow optimized for simultaneous computation of node embedding, edge embedding, and message passing, which is generally applicable to all models. We also propose a rich library of model-specific components. Second, we deliver ultra-fast real-time GNN inference without any graph pre-processing, making it agnostic to dynamically changing graph structures. Third, we verify our architecture on the Xilinx Alveo U50 FPGA board and measure the on-board end-to-end performance. We achieve a speed-up of up to 24-254x against CPU (6226R) and 1.3-477x against GPU (A6000) (with batch sizes 1 through 1024); we also outperform the SOTA GNN accelerator I-GCN by 1.26x speedup and 1.55x energy efficiency over four datasets. Our implementation code and on-board measurement are publicly available on GitHub., 13 pages, 10 figures. Accepted at HPCA 2023

Background Repairing radiation-induced bone injuries remains a significant challenge in the clinic, and few effective medicines are currently available. Psoralen is a principal bioactive component of Cullen corylifolium (L.) Medik and has been reported to have antitumor, anti-inflammatory, and pro-osteogenesis activities. However, less information is available regarding the role of psoralen in the treatment of radiation-induced bone injury. In this study, we explored the modulatory effects of psoralen on skeletal stem cells and their protective effects on radiation-induced bone injuries. Methods The protective effects of psoralen on radiation-induced osteoporosis and irradiated bone defects were evaluated by microCT and pathological analysis. In addition, the cell proliferation, osteogenesis, and self-renewal of SSCs were explored. Further, the underlying mechanisms of the protective of psoralen were investigated by using RNA sequencing and functional gain and loss experiments in vitro and in vivo. Statistical significance was analyzed using Student's t test. The one-way ANOVA was used in multiple group data analysis. Results Here, we demonstrated that psoralen, a natural herbal extract, mitigated radiation-induced bone injury (irradiation-induced osteoporosis and irradiated bone defects) in mice partially by rescuing the stemness of irradiated skeletal stem cells. Mechanistically, psoralen restored the stemness of skeletal stem cells by alleviating the radiation-induced suppression of AKT/GSK-3β and elevating NRF2 expression in skeletal stem cells. Furthermore, the expression of KEAP1 in skeletal stem cells did not significantly change in the presence of psoralen. Moreover, blockade of NRF2 in vivo partially abolished the promising effects of psoralen in a murine model of irradiation-induced osteoporosis and irradiated bone regeneration. Conclusions In summary, our findings identified psoralen as a potential medicine to mitigate bone radiation injury. In addition, skeletal stem cells and AKT-GSK-3β and NRF2 may thus represent therapeutic targets for treating radiation-induced bone injury. Graphical Abstract

Agile hardware development requires fast and accurate circuit quality evaluation from early design stages. Existing work of high-level synthesis (HLS) performance prediction usually needs extensive feature engineering after the synthesis process. To expedite circuit evaluation from as earlier design stage as possible, we propose a rapid and accurate performance modeling, exploiting the representation power of graph neural networks (GNNs) by representing C/C++ programs as graphs. The contribution of this work is three-fold. First, we build a standard benchmark containing 40k C synthesizable programs, which includes both synthetic programs and three sets of real-world HLS benchmarks. Each program is implemented on FPGA to generate ground-truth performance metrics. Second, we formally formulate the HLS performance prediction problem on graphs, and propose multiple modeling strategies with GNNs that leverage different trade-offs between prediction timeliness (early/late prediction) and accuracy. Third, we further propose a novel hierarchical GNN that does not sacrifice timeliness but largely improves prediction accuracy, significantly outperforming HLS tools. We apply extensive evaluations for both synthetic and unseen real-case programs; our proposed predictor largely outperforms HLS by up to 40X and excels existing predictors by 2X to 5X in terms of resource usage and timing prediction.

Groundwater is regularly used for many purposes, such as drinking and agricultural irrigation systems. Still, it contains high levels of radionuclides (e.g.

Scenarios generation is a critical part in planning and operation in high renewable energy penetratied power systems. However, the statistical assumptions of traditional parametric methods may not hold for all types of wind farms. In this paper, a data-driven artificial intelligence approach is presented to generate wind power output scenarios based on generative adversarial networks (GANs). Unlike traditional probabilistic model-based techniques which are typically difficult to scale or sample, the proposed method is data-driven and captures patterns of wind power generation. First, the GAN network structure is constructed, and the Wasserstein distance is employed as the discriminator’s loss function. The GAN training then enables the generator to learn random noise and actual history data. Finally, the scenario generation approach based on Monte Carlo simulation and GANs are compared. It shows that the scenarios generated by proposed method can accurately describe the uncertainty of wind power output.

Despite the great success of High-Level Synthesis (HLS) tools, we observe several unresolved challenges: 1) the high-level abstraction of programming styles in HLS conceals optimization opportunities; 2) existing HLS tools do not provide flexible trade-offs among different objectives and constraints; 3) the actual quality of the resulting RTL designs is hard to predict. To this end, we propose an end-to-end framework, IRONMAN. The primary goal is to enable a flexible and automated design space exploration (DSE), which can provide either optimized solutions under user-specified constraints, or Pareto trade-offs among different objectives (e.g., resource types, area, and latency). IronMan consists of three components: GPP (a highly accurate graph-neural-network-based performance predictor), RLMD (a reinforcement-learning-based DSE engine that explores the optimized resource allocation strategy), and CT (a code transformer that assists RLMD and GPP by extracting data flow graphs from original HLS C/C++). Experimental results show that, 1) GPP achieves high prediction accuracy, reducing prediction errors of HLS tools by 10.9X in resource usage and 5.7X in timing; 2) RLMD obtains optimized or Pareto solutions outperforming genetic algorithm and simulated annealing by 12.7% and 12.9%, respectively; 3) IronMan can find optimized solutions perfectly matching various DSP constraints, with 2.54X fewer DSPs and up to 6X shorter latency than those of HLS tools. IronMan is also up to 400X faster than meta-heuristic techniques and HLS tools.