Low-light image enhancement (LLIE) aims at refining illumination and restoring the detail of low-light image. Current deep LLIE models still face two issues: low-quality detail-recovery result due to information loss, and complex model design. On one hand, current methods usually adopt U-Net with multiple feature scaling operations as main structure, but feature scaling inevitably loses informative visual primitives, which will result in blur textures and inaccurate illumination. On the other hand, current models are usually complicated and even redundant, which goes against the original goal of building plain model to effectively handle the LLIE task. To address these issues, we branch out to propose a simple yet effective deep LLIE architecture, termed Full-Resolution Context Network (FRC-Net). Specifically, to avoid information loss caused by feature scaling, we propose a novel full-resolution representation strategy to replace all feature scaling operations. The model structure of FRC-Net is very simple, which only contains 12 cascaded layers: 7 convolution layers and 5 newly-designed context attention (CA) modules. CA module is mainly designed to overcome the limited receptive field caused by shallow structures, by not only learning global context but also retaining local details. Extensive experiments show that our FRC-Net obtains better detail-recovery quality, which performs favorably against the current SOTA methods.

Autonomous vehicle requires a general planner for all possible scenarios. Existing researches design such a planner by a unified scenario description. However, it may significantly increase the planner complexity even in some simple tasks, e.g., car following, further resulting in unsatisfactory driving performance. This work aims to design a general planner which can 1) drive in all possible scenarios and 2) have lower complexity in some common scenarios. To this end, this work proposes a pertinent boundary for multi-scenario driving planning. The total approach is named as Pertinent Boundary-based Unified Decision system. Based on the original drivable area, the pertinent boundary can further support motion status and semantics of the traffic elements, which provides the potential of pertinent performance for given scenarios. The pertinent boundary can support unified driving with the drivable area, in the meantime, can be pertinently modified to support the pertinent driving decisions for identified driving scenarios (e.g., car-following, junction left turning). It will further avoid the bump between the connections of the scenarios due to the continuity of space boundary. Thus, the planner is suitable for the fully autonomous driving. The proposed method is validated in different classical driving decision scenarios. Results show that the proposed method can support pertinent driving decisions in identified scenarios, in the meantime, assure generalized cross-scenario planning when no scenario information is available. Such a method shed light on fully autonomous driving by pertinence improvement of multi-scenario decision in the complex real world.

An aberration correction algorithm has been implemented and demonstrated in an echocardiographic clinical trial using 2D imaging. The method estimates and compensates arrival time errors between different sub- aperture processor signals in a matrix array probe in post processing. Five standard views of channel data cineloops were recorded from 22 patients. Using a coherence metric, the aberration correction algorithm was shown to improve image quality in all 7380 processed frames. In a blinded and left-right-randomized side-by-side evaluation, four cardiologists (two experienced and two in training) preferred the aberration corrected image in 97% of the cases. The feedback from the clinicians was that the images appeared sharper with better contrast and less noise. Many structures like valve leaflets, chordae, endocardium, and endocardial borders appeared narrower and more clearly defined in the aberration corrected images. An important finding is that aberration correction improves contrast between the endocardium and ventricle cavities for all processed images. This was confirmed by the cardiologists in their feedback, and quantified with a median global gain difference estimate between the aberration-corrected and non- corrected images of 1.2 dB.

Cine cardiac magnetic resonance (CMR) imaging is considered the gold standard for cardiac function evaluation. However, cine CMR acquisition is inherently slow and in recent decades considerable effort has been put into accelerating scan times without compromising image quality or the accuracy of derived results. In this paper, we present a fully-automated, quality-controlled integrated framework for reconstruction, segmentation and downstream analysis of undersampled cine CMR data. The framework enables active acquisition of radial k-space data, in which acquisition can be stopped as soon as acquired data are sufficient to produce high quality reconstructions and segmentations. This results in reduced scan times and automated analysis, enabling robust and accurate estimation of functional biomarkers. To demonstrate the feasibility of the proposed approach, we perform realistic simulations of radial k-space acquisitions on a dataset of subjects from the UK Biobank and present results on in-vivo cine CMR k-space data collected from healthy subjects. The results demonstrate that our method can produce quality-controlled images in a mean scan time reduced from 12 to 4 seconds per slice, and that image quality is sufficient to allow clinically relevant parameters to be automatically estimated to within 5% mean absolute difference.

Abstract:  Recent years have witnessed an exponential growth of the commercial space industry, including rocket launch, satellite network deployment, private space travel, and even extraterrestrial colonization. Several trends are predicted in this unprecedented transition to an era of space-enabled broadband access. Comments: To appear in Computer magazine, IEEE Related DOI: 10.1109/MC.2022.3160472

This work considers optimizing the IRS elements' selection process to reduce the average BER. In addition to BER reduction, reducing the number of elements reduces the overhead associated with phase information feedback to the IRS panel, and allows serving other users in the network. Moreover, the phase error distribution is derived for the case of LoS communications in closed-form.

This work considers the problem of joint IRS elements and power allocation for UAV networks in the presence of imperfect phase compensation and energy constraints. The allocation problem is formulated to maximize the network sum-rate with minimum individual rate and energy constraints.

The integration between the satellite network and the terrestrial network will play a key role in the upcoming sixth-generation (6G) of mobile cellular networks thanks to the wide coverage and bandwidth offered by satellite networks. To leverage this integration, we propose a proactive traffic offloading scheme in an integrated multi-satellite terrestrial network (IMSTN) that considers the future networks’ heterogeneity and predicts their variability. Our proposed offloading scheme hinges on traffic prediction to answer the stringent requirements of data-rate, latency and reliability imposed by heterogeneous and coexisting services and traffic namely enhanced mobile broadband (eMBB), massive machine-type communications (mMTC) and ultra-reliable low latency communication (URLLC). However, the fulfillment of these requirements during offloading in  dynamic IMSTN comes at the expense of significant energy consumption and introduces inherently supplementary latency. Therefore, our offloading scheme aims to balance the fundamental trade-offs first between energy consumption and the achievable data-rate and second between energy consumption and latency while meeting the respective needs of the present traffic.

The balance between detailed technical description, representation of uncertainty and computational complexity is central in long-term scheduling models applied to hydro-dominated power system. The aggregation of complex hydropower systems into equivalent energy representations (EER) is a commonly used technique to reduce dimensionality and computation time in scheduling models. This work presents a method for coordinating the EERs with their detailed hydropower system representation within a model based on stochastic dual dynamic programming (SDDP). SDDP is applied to an EER representation of the hydropower system, where feasibility cuts derived from optimization of the detailed hydropower are used to constrain the flexibility of the EERs. These cuts can be computed either before or during the execution of the SDDP algorithm and allow system details to be captured within the SDDP strategies without compromising the convergence rate and state-space dimensionality. Results in terms of computational performance and system operation are reported from a test system comprising realistic hydropower watercourses.

The project is focusing on building a framework of IoT security. Our main goals are: to model IoT attacks and defences in an emulated IoT networks; and to evaluate the impact of different defence techniques on IoT attacks by using multiple security metrics. \tightlist This framework can create IoT networks with different settings and configurations on IoT devices and network topologies, model malware attacks against the emulated IoT networks and evaluate relevant defences. The simulation software is built with the aim to offer flexibility and adaptability to users. Also, it can be extended with new features and functionalities to simulate recent discovered malware attacks and defences.

Purpose: To compare the diagnostic accuracy and explainability of a new Vision Transformer deep learning technique, Data-efficient image Transformer (DeiT), and Resnet-50, trained on fundus photographs from the Ocular Hypertension Treatment Study (OHTS) to detect primary open-angle glaucoma (POAG) and to identify the salient areas of the photographs most important for each model’s decision-making process. Study Design: Evaluation of a diagnostic technology Subjects, Participants, and/or Controls: 66,715 photographs from 1,636 OHTS participants and an additional five external datasets of 16137 photographs of healthy and glaucoma eyes. Methods, Intervention, or Testing: DeiT models were trained to detect five ground truth OHTS POAG classifications: OHTS Endpoint Committee POAG determinations due to disc changes (Model 1), visual field changes (Model 2), or either disc or visual field changes (Model 3) and reading center determinations based on disc (Model 4) and visual fields (Model 5). The best-performing DeiT models were compared to ResNet-50 on OHTS and five external datasets. Main Outcome Measures: Diagnostic performance was compared using areas under the receiver operating characteristic curve (AUROC) and sensitivities at fixed specificities. The explainability of the DeiT and ResNet-50 models was compared by evaluating the attention maps derived directly from DeiT to 3 gradient-weighted class activation map generation strategies. Results: Compared to our best-performing ResNet-50 models, the DeiT models demonstrated similar performance on the OHTS test sets for all five-ground truth POAG labels; AUROC ranged from 0.82 (Model 5) to 0.91 (Model 1). However, the AUROC of DeiT was consistently higher than ResNet-50 on the five external datasets. For example, AUROC for the main OHTS endpoint (Model 3) was between 0.08 and 0.20 higher in the DeiT compared to ResNet-50 models. The saliency maps from the DeiT highlight localized areas of the neuroretinal rim, suggesting the use of important clinical features for classification, while the same maps in the ResNet-50 models show a more diffuse, generalized distribution around the optic disc, Conclusions: Vision transformer has the potential to improve the generalizability and explainability of deep learning models for the detection of eye disease and possibly other medical conditions that rely on imaging modalities for clinical diagnosis and management.

A Game for Task Offloading in Reputation-Based Consortium Blockchain Networks

Antenna array mutual impedances are measured from embedded element patterns. To do this practically, additional relationships are developed for Thevenin-driven embedded element patterns where the driven element has a different source impedance from the non-driven elements.

This paper analyzes the performance of an intelligent reflecting surface (IRS)-aided short-packet communication system over Nakagami-m fading channels. In this system, the base station adopts non-orthogonal multiple access (NOMA) to transmit to a near user and a far user, and its transmission to the far user is aided by the IRS. Both continuous phase shifts (CPSs) and discrete phase shifts (DPSs) are considered for the IRS. By using order statistic and moment matching, we approximate the end-to-end channel gain to follow the Gamma distribution and then derive the approximate closed-form expressions for the average block error rate (BLER) in the high signal-to-noise ratio (SNR) regime and then conduct a refifined analysis of the interplay among the minimum common blocklength, optimal power allocation, and the related parameters of IRS (e.g., the number of elements of the IRS) for achieving target BLERs at two users. Finally, numerical and simulation results demonstrate the accuracy of theoretical analysis. They also show that the IRS-aided short-packet NOMA system outperforms the system without IRS, and that DPSs with 3-bit quantization achieve a similar performance as the optimal CPSs.

In hardware such as FPGAs, Kenneth Batcher’s Odd-Even Merge Sort and Bitonic Merge Sort are the state-of-the-art methodologies used to quickly sort a list of more than 16 input values. Both sorting networks feature merges of 2 sorted input lists into a single sorted output list. For both, a full sort of 64 and 512 input values requires 21 and 45 serial stages, respectively. Multiway merge sorting networks described here require significantly fewer serial stages. For example, 8-way merge networks fully sort 64 and 512 input values in 9 and 20 serial stages, less than half the number of the respective 2-way networks. When the multiway merge sorting networks utilize the single-stage N-sorters recently defined by the authors, they are considerably faster than Batcher’s networks. In the AMD-Xilinx Ultrascale+ xcvu9p FPGA, the two 8-way merge networks have speedups of 1.85 and 1.74 versus the comparable 2-way networks. A fully pipelined 3-way merge network in this FPGA is capable of fully sorting 500 million lists of 729 unsorted 32-bit values in one second. In software, multiway merge methods are used to find the median of certain pixel rectangles in images, since the median can be determined in fewer stages than are required to fully sort the rectangle. However, the software still requires a series of many 2-sorter operations to find a median. These multiway merge median methods are dramatically sped up in hardware, where the authors’ new single-stage N-sorters and N-filters operate in parallel in each stage of the merge process.

Abstract:  In this paper, we propose two new provable algorithms for tracking online low-rank approximations of high-order streaming tensors with missing data. The first algorithm, dubbed adaptive Tucker decomposition (ATD), minimizes a weighted recursive least-squares cost function to obtain the tensor factors and the core tensor in an efficient way, thanks to the alternating minimization framework and the randomized sketching technique. Under the Canonical Polyadic (CP) model, the second algorithm called ACP is developed as a variant of ATD when the core tensor is imposed to be identity. Both algorithms are low-complexity tensor trackers that have fast convergence and low memory storage requirements. A unified convergence analysis is presented for ATD and ACP to justify their performance. Experiments indicate that the two proposed algorithms are capable of streaming tensor decomposition with competitive performance with respect to estimation accuracy and runtime on both synthetic and real data. Code: https://github.com/thanhtbt/tensor_tracking Comment: to appear in Elsevier Patterns

In this paper, we apply and evaluate several machine learning and deep learning methods, along with various feature extraction and word-embedding techniques, on a consolidated dataset of 20600 instances, for hate speech detection from tweets and comments in Hinglish. The experimental results reveal that deep learning models perform better than machine learning models in general. Among the deep learning models, the CNN-BiLSTM model with word2vec word embedding provides the best results. The model yields 0.876 accuracy, 0.830 precision, 0.840 recall and 0.835 F1-score. These results surpass the recent state-of-art approaches.

Since its introduction with Bitcoin in 2009, blockchain technology has received tremendous attention by academia, industry, politics and media alike, in particular, through extended blockchain-based visions such as smart contracts, decentralized finance, and, most recently, Web3. The critical prerequisite for any such blockchain-based vision to be turned into reality is uncapped scalability. Furthermore, and equally important, blockchain technology needs to transcend the stage of specialized tokens into an adaptive, heterogeneous tokenization platform. In this paper, we explain the Alphabill family of technologies that addresses both unlimited scalability and unrestricted adaptivity. We deliver a sharded blockchain technology with unlimited scalability and performance, called KSI Cash, which is based on a new form of electronic money scheme, the bill scheme. We present performance tests of KSI Cash that we have conducted with the European Central Bank and a group of eight national central banks from the Eurosystem in order to assess the technological feasibility of a digital euro, showing the system operating with 100 million wallets and 15 thousand transactions per second (under simulation of realistic usage), having an estimated carbon footprint of 0.0001g CO2 per transaction (Bitcoin = 100 kg and more); furthermore, showing the system operating with up to 2 million payment orders per second, an equivalent of more than 300.000 transactions per second (in a laboratory setting with the central components of KSI Cash), scaling linearly in terms of the number of deployed shards.  We explain, in detail, the key concepts that unlock this performance (i.e., the concepts of the bill money scheme). The results provide evidence that the scalability of our technology is unlimited in both permissioned and permissionless scenarios, resulting into the Alphabill Money technology. Next, we contribute the architecture of a universal tokenization platform that allows for universal asset tokenization, transfer and exchange as a global medium of exchange,  called Alphabill platform. We reveal the crucial conceptual and technical contributions of the platform’s architecture and their interplay, including the data structures of KSI Cash and Alphabill Money, the dust collection solution of Alphabill Money, and the  atomic swap solution of the Alphabill platform.