Applied Technologies (United States)

Neural Network processing requires extensive hardware resources and computation power, hence data is captured and transmitted to the cloud for AI inference. With extensive deployment of smart city systems, reliance on Convolutional Neural Network (CNN) based object detection algorithms like YOLO for real-time object detection and classification have also increased resulting in enhanced requirement for computing power. Keeping this situation in mind, a simple and generalized convolution accelerator architecture is proposed that is optimized for edge deployment, providing high FPS, as well as saving energy and compute costs associated with data transmission and cloud computing. The accelerator can work with 16-bit floating and fixed-point representations and has a relatively simple control scheme. The 16 bit representation results in a highly accurate inference, with little to no loss in accuracy. The proposed architecture has been designed specifically for implementation in edge devices and has been tested/validated on AMD Xilinx ZCU106 FPGA development board, with performance of 614.1 GFLOPs/sec at 150MHz, nearly 2.4 times more than the required 258 GFLOPs/sec in YOLOv8x.

Cross-electrophile coupling (XEC) reactions that forge C(sp2)─C(sp3) bonds have received considerable attention due to the vast libraries of commercially available organohalides. However, the incorporation of ubiquitous aryl chlorides remains a challenge due to the slow rate of oxidative addition of metal catalysts to these electrophiles relative to iodide and bromide derivatives. This study reports a simple metallaphotoredox-catalyzed C(sp2)─C(sp3) cross-electrophile coupling (XEC) method for the selective coupling of a broad range of aryl chlorides and alkyl halides. By design, this methodology exploits the enabling interplay of a bidentate 2-(1H-imidazol-2-yl)pyridine ligand, incorporating the strong σ-donor capacity of a 1H-imidazole moiety with the moderate π-acceptor capacity of a pyridine unit. The synergy provided by these modular components allowed for the electronic requirements of elementary steps to be accommodated, such as the synchronous integration of alkyl radical generation (typically relatively fast) and oxidative addition of low valent nickel species to aryl chlorides (often relatively slow). This protocol enabled the efficient late-stage diversification of the drugs Me-fenofibrate, indomethacin, and etoricoxib to enhance their sp3-rich complexity. Using an equimolar ratio of substrates, this methodology facilitated a decagram XEC in continuous flow, showcasing the applicability of the method for large scale synthesis.

Abstract Variational quantum circuits (VQC) lie at the forefront of quantum machine learning research. Still, the use of quantum networks for real data processing remains challenging as the number of available qubits cannot accommodate a large dimensionality of data—if the usual angle encoding scenario is used. To achieve dimensionality reduction, Principal Component Analysis is routinely applied as a pre-processing method before the embedding of the classical features on qubits. In this work, we propose an alternative method which reduces the dimensionality of a given dataset by taking into account the specific quantum embedding that comes after. This method aspires to make quantum machine learning with VQCs more versatile and effective on datasets of high dimension. At a second step, we propose a quantum-inspired classical autoencoder model which can be used to encode information in low latent spaces. The power of our proposed models is exhibited via numerical tests. We show that our targeted dimensionality reduction method considerably boosts VQC’s performance, and we also identify cases for which the second model outperforms classical autoencoders in terms of reconstruction loss.

A paucity of studies investigates the outcomes of flap reconstruction in lower extremity acute burns. The aim of this study is to report outcomes of lower extremity acute burn requiring pedicled or free flap coverage. A retrospective cohort study was conducted to compare the outcomes of patients undergone pedicled versus free flap reconstruction of acute lower extremity burns, between August 2010 and December 2022. Collected data included demographics, injury and flap characteristics, complications, and reoperations. χ 2 tests were used to measure differences in complication rates between pedicled and free flaps. A total of 28 patients were involved in the study. Among them, 17 patients underwent 28 pedicled flap procedures, while 11 patients received a single free flap surgery each. In the free flap group, the overall complication rate was 54.5%. In the pedicled flap group, the overall complication rate was 25.0%. Free flaps showed a significantly higher rate of total flap loss compared to pedicled flaps (18.2% vs 0%, P = 0.021). Other differences were not statistically significant. Flap coverage in lower extremity acute burns is rarely employed. Yet, in case of critical structures exposure it is often necessary. However, it is important to be aware of the high risk of complications, especially for more complex reconstructions requiring free tissue transfer.

Unmanned Aerial Vehicles (UAVs) encounter wind gusts during outdoor operations, impacting their position holding, particularly for quadrotors. This vulnerability is amplified during the autonomous docking to outdoor charging stations. The integration of real-time wind preview information for UAV gust rejection control has become more feasible with advances in remote wind sensor technologies like LiDAR. In this study, a ground-based LiDAR system is proposed to predict wind gusts at the landing site of quadrotors. The acquired wind preview data are subsequently utilized by the Model Predictive Control (MPC) to effectively mitigate disturbances. To validate the proposed methodology, a nonlinear simulation environment has been established using LiDAR data collected from comprehensive field tests. The results demonstrate a notable improvement in the system performance compared to benchmark results. This research underscores the practical utility of real-time wind preview information, facilitated by LiDAR technology, in enhancing the overall operational resilience of UAVs, especially quadrotors, during challenging environmental conditions.