Search

Capturing fine spatial, spectral, and temporal information of the scene is highly desirable in many applications. However, recording data of such high dimensionality requires significant transmission bandwidth. Current computational imaging methods can partially address this challenge but are still limited in reducing input data throughput. In this paper, we report a video-rate hyperspectral imager based on a single-pixel photodetector which can achieve high-throughput hyperspectral video recording at a low bandwidth. We leverage the insight that 4-dimensional (4D) hyperspectral videos are considerably more compressible than 2D grayscale images. We propose a joint spatial-spectral capturing scheme encoding the scene into highly compressed measurements and obtaining temporal correlation at the same time. Furthermore, we propose a reconstruction method relying on a signal sparsity model in 4D space and a deep learning reconstruction approach greatly accelerating reconstruction. We demonstrate reconstruction of 128 × 128 hyperspectral images with 64 spectral bands at more than 4 frames per second offering a 900× data throughput compared to conventional imaging, which we believe is a first-of-its kind of a single-pixel-based hyperspectral imager. The authors showcase a video-rate hyperspectral imager based on a single-pixel photodetector that can achieve high-throughput hyperspectral video recording at a low bandwidth. Specifically, they propose a joint spatial-spectral encoding scheme which can encode the scene into highly compressed single-pixel measurements and obtain temporal correlation at the same time. [ABSTRACT FROM AUTHOR]

The COVID-19 pandemic has caused significant harm globally, prompting us to prioritize prevention measures. Effective hand-washing is one of the most critical and straightforward measures that can help prevent the spread of this virus. Medical staff's hands are considered a major source of hospital infection. Effective hand-washing can prevent up to 30% of diarrhea-related illnesses, which is crucial in preventing nosocomial infections (Tartari et al. in Clin Microbiol Infect 23(9):596–598, 2017). This paper proposes an electronic-based real-time hand-washing identification framework called Alpha Hand Washing (ALPHA HW). The system uses camera-based identification, edge computing, and deep learning to automatically identify correct hand-washing behaviors, thereby facilitating effective hand-washing (Bertasius et al. in: Computer vision and pattern recognition, 2015). We achieved an accuracy of 78.0% mAP and a speed of 52 FPS in detecting scenes using specific monitoring datasets in hospitals by constructing the complex recognition system into a grid computing problem. Leveraging edge computing, our system achieves real-time identification with low memory consumption and high-efficiency computation. Alpha HW presents scientific and financial values in epidemic prevention and control that can facilitate popularization to reduce virus spread (Bewley et al. in 2016 IEEE international conference on image processing, 2016). [ABSTRACT FROM AUTHOR]

Posterolateral (PL)-depression fractures of the tibial plateau are difficult to manage. The aim of this study was: (1) to present our experience with a novel technique of robot-assisted percutaneous reduction and fixation and (2) to compare it with the traditional percutaneous screw osteosynthesis (PSO) technique for the treatment of PL-depression tibial plateau fractures. The clinical data of patients with PL-depression tibial plateau fractures treated by robot-assisted percutaneous reduction and fixation technique and traditional PSO technique from January 2017 to January 2021 were retrospectively analyzed. Among them, there were 18 cases in the robot-assisted group (RA group) and 23 cases in the PSO group. All fractures were unilateral, closed and fresh PL-depression fractures of the tibial plateau. Patients in the RA group were treated by robot assisted reduction and minimally invasive absorbable screw fixation. The PSO group was treated by closed reduction and percutaneous absorbable screw osteosynthesis. The incision length, operation time, intraoperative blood loss, fluoroscopic times, inpatient time, weight training time and postoperative complications of the two groups were statistically analyzed. The Rasmussen radiological score was used to assess the reduction quality after operation while the Rasmussen functional score was used to evaluate knee joint functions at one year postoperatively. All patients were followed for at least one year. There was no significant difference in demographic information between the two groups (p > 0.05). Compared with the PSO group, the RA group showed less fluoroscopic times and better Rasmussen radiological and functional scores (p < 0.05). Besides, there was no significant difference in the incidence of postoperative complications between the two groups (p > 0.05). The novel robot-assisted percutaneous reduction and fixation technique had the characteristics of less radiation, accurate reduction and fixation. It could accelerate the rehabilitation of patients with PL-depression fractures of the tibial plateau and enable patients to obtain good joint functions. [ABSTRACT FROM AUTHOR]

The Chasing All Transients Constellation Hunters (CATCH) space mission plans to launch three types of micro-satellites (A, B, and C). The type-B CATCH satellites are dedicated to locating transients and detecting their time-dependent energy spectra. A type-B satellite is equipped with lightweight Wolter-I X-ray optics and an array of position-sensitive multi-pixel Silicon Drift Detectors. To optimize the scientific payloads for operating properly in orbit and performing the observations with high sensitivities, this work performs an in-orbit background simulation of a type-B CATCH satellite using the Geant4 toolkit. It shows that the persistent background is dominated by the cosmic X-ray diffuse background and the cosmic-ray protons. The dynamic background is also estimated considering trapped charged particles in the radiation belts and low-energy charged particles near the geomagnetic equator, which is dominated by the incident electrons outside the aperture. The simulated persistent background within the focal spot is used to estimate the observation sensitivity, i.e. 4.22 × 10 - 13 erg cm - 2 s - 1 with an exposure of 10 4 s and a Crab-like source spectrum, which can be utilized further to optimize the shielding design. The simulated in-orbit background also suggests that the magnetic diverter just underneath the optics may be unnecessary in this kind of micro-satellites, because the dynamic background induced by charged particles outside the aperture is around 3 orders of magnitude larger than that inside the aperture. [ABSTRACT FROM AUTHOR]

Stochastic gradient methods (SGMs) are the predominant approaches to train deep learning models. The adaptive versions (e.g., Adam and AMSGrad) have been extensively used in practice, partly because they achieve faster convergence than the non-adaptive versions while incurring little overhead. On the other hand, asynchronous (async) parallel computing has exhibited significantly higher speed-up over its synchronous (sync) counterpart. Async-parallel non-adaptive SGMs have been well studied in the literature from the perspectives of both theory and practical performance. Adaptive SGMs can also be implemented without much difficulty in an async-parallel way. However, to the best of our knowledge, no theoretical result of async-parallel adaptive SGMs has been established. The difficulty for analyzing adaptive SGMs with async updates originates from the second moment term. In this paper, we propose an async-parallel adaptive SGM based on AMSGrad. We show that the proposed method inherits the convergence guarantee of AMSGrad for both convex and non-convex problems, if the staleness (also called delay) caused by asynchrony is bounded. Our convergence rate results indicate a nearly linear parallelization speed-up if τ = o (K 1 4) , where τ is the staleness and K is the number of iterations. The proposed method is tested on both convex and non-convex machine learning problems, and the numerical results demonstrate its clear advantages over the sync counterpart and the async-parallel nonadaptive SGM. Our code has been released at https://github.com/RPI-OPT/APAM. [ABSTRACT FROM AUTHOR]

China, as the world's largest energy consumer, has made the green economy a central component of its economic development strategy. However, how to effectively play the government's crucial role in promoting the development of the green economy has become the focus of research by a significant number of economic experts. This paper uses the Super-SBM model to measure the green economy development index by introducing carbon dioxide emissions and industrial "three wastes" emissions and analyzes the relationship between fiscal decentralization, green technology innovation, and the green economy from the vantage point of local government behavior. It is discovered that fiscal decentralization significantly inhibits the development of the green economy, and local green technology innovation activities in the last period will amplify this negative impact. The above findings pass the robustness test. After introducing comparative analysis of economic growth indicators that are measured by the stochastic frontier analysis (SFA), the results show that only in the eastern region does fiscal decentralization both drive economic growth and do not inhibit green economy development by local government officials' political promotion motives and self-interested preferred expenditures, but overall economic promotion and green economy inhibition caused by fiscal decentralization exist simultaneously in the Yangtze River Economic Belt region, and significant heterogeneity differences exist in the rest of the regions. The findings suggest that regulating local government fiscal behavior and improving fiscal transparency are very important to promote the development of China's green economy. [ABSTRACT FROM AUTHOR]

Stochastic gradient methods (SGMs) have been extensively used for solving stochastic problems or large-scale machine learning problems. Recent works employ various techniques to improve the convergence rate of SGMs for both convex and nonconvex cases. Most of them require a large number of samples in some or all iterations of the improved SGMs. In this paper, we propose a new SGM, named PStorm, for solving nonconvex nonsmooth stochastic problems. With a momentum-based variance reduction technique, PStorm can achieve the optimal complexity result O (ε - 3) to produce a stochastic ε -stationary solution, if a mean-squared smoothness condition holds. Different from existing optimal methods, PStorm can achieve the O (ε - 3) result by using only one or O(1) samples in every update. With this property, PStorm can be applied to online learning problems that favor real-time decisions based on one or O(1) new observations. In addition, for large-scale machine learning problems, PStorm can generalize better by small-batch training than other optimal methods that require large-batch training and the vanilla SGM, as we demonstrate on training a sparse fully-connected neural network and a sparse convolutional neural network. [ABSTRACT FROM AUTHOR]

Purpose: To assess the efficacy of autogenous "structured" bone grafting (ASBG), it was combined with superior plate (SP) revision operations for recalcitrant clavicular midshaft aseptic nonunion (CMAN).Methods: This retrospective study included 12 patients who suffered from failure of autologous cancellous bone grafting (ACBG) and SP fixation because of CMAN. Visual analogue scale (VAS) data for pain and disabilities of arm, shoulder, and hand (DASH) scores of patients who underwent these procedures from January 2019 to December 2020, obtained before surgery and at the final follow-up time, were analysed.Results: The average time between primitive fracture and this operative treatment was 29 months (15-38 months). The average duration of surgery was 153 minutes (range, 115-230 min), and the average blood loss was 560 ml (range, 350-860 ml). Complications occurred in two cases (16.67%): one was persistent pain at the donor site, and the other was a calf muscle vein thrombosis. No tissue infection was observed during follow-up. The mean follow-up time was 18 months (range, 12-30 months). All fractures progressed to osseous healing at a mean time of 14 weeks (range, 12-16 weeks). The mean pain VAS score significantly improved, from 4.8 ± 1.7 pre-operatively to 1.9 ± 1.1 at the final follow-up (P = 0.01). The mean DASH score improved significantly from 30.1 ± 11.2 pre-operatively to 7.8 ± 4. 2 at the final follow-up (P < 0.01).Conclusions: ASBG combined with SP revision surgery achieved excellent clinical outcomes in patients with recalcitrant CMAN. [ABSTRACT FROM AUTHOR]

In the present study, we presented our experience with a new modified technique of shortening and re-lengthening using a monolateral external frame combined with a calcaneal intramedullary nail and compared it with the bone transport technique for the treatment of distal tibial periarticular post-traumatic defects. Forty-one patients were retrospectively analyzed. Among them 19 were treated using our modified shortening and re-lengthening technique (MSR group) and 22 by bone transport (BT group). The difference in external fixation time (EFT), external fixation index (EFI), self-rating anxiety scale (SAS), and complications were compared between the two groups. The mean EFT was 3.4 ± 0.6 months in the MSR group and 7.5 ± 1.4 months in the BT group; the EFI was 0.57 ± 0.06 month/cm and 1.32 ± 0.23 month/cm, respectively. The EFT, EFI, and SAS scores were significantly lower in the MSR group than in the BT Group (p < 0.05). The mean number of complications per patient in the BT group was nearly 2.4 times that of the MSR group (p < 0.05). Our modified shortening and re-lengthening technique reduced the EFI and complication incidence compared to the bone transport technique. Therefore, patients with distal tibial periarticular post-traumatic defects can achieve great satisfaction with this new technique. [ABSTRACT FROM AUTHOR]

With the extensive application of machine vision in Agriculture, plant recognition is becoming an important research territory. Due to illumination change, occlusion problem, green background and other factors, the image segmentation quality of plant is uneven. When there are significantly different colors, such as red, green and blue, between plant target and background, classical image processing methods are up to the task. However, in near color scene, for example dark green target and bright green background, plant target recognition is still a very challenging task. To segment plants in above scene, the near color recognition method based on residual vector and SVM has been developed. Firstly, the color vectors were projected to the plane that crosses the point Origin(0, 0, 0) and is perpendicular to reference color vector B0(255, 255, 255). After projection, the significantly different color vectors were distributed in different polar angle ranges, while the near color vectors were concentrated in the same polar angle range. For near color vectors, the small polar angle difference, namely roll, was regarded as redundant information. Then, the angle θ between target color vector A(r, g, b) and B0, along with the norm of A, namely A → was calculated. As a result, the three-dimensional target color vector A was converted to the two-dimensional residual vector R A → θ . Finally, SVM classifier is used to identify the residual vector. The results show that the linear recognition rate is 90.25%, the average recognition speed 0.243 s, the nonlinear recognition rate 87.47%, and the average recognition speed 0.254 s. This study provides theoretical reference for plant target recognition in the near color scene. [ABSTRACT FROM AUTHOR]

Convex hulls of monomials have been widely studied in the literature, and monomial convexifications are implemented in global optimization software for relaxing polynomials. However, there has been no study of the error in the global optimum from such approaches. We give bounds on the worst-case error for convexifying a monomial over subsets of. This implies additive error bounds for relaxing a polynomial optimization problem by convexifying each monomial separately. Our main error bounds depend primarily on the degree of the monomial, making them easy to compute. Since monomial convexification studies depend on the bounds on the associated variables, in the second part, we conduct an error analysis for a multilinear monomial over two different types of box constraints. As part of this analysis, we also derive the convex hull of a multilinear monomial over. [ABSTRACT FROM AUTHOR]