Near-infrared transmission through the finger can capture the vein structure for identity recognition. However, in outdoor applications, finger vein imaging is significantly affected by environmental illumination resulting in low recognition performance. Existing methods typically address this issue by constructing multi-illumination models, but collecting multi-illumination images from individual is challenging, and overexposure can cause venous structure distortion. This paper proposes MDA-Net, a Multi-illumination Domain Adaptive Network for finger vein recognition, which is engineered to excel in the dynamic outdoor lighting landscape with various conditions including overexposure, using only data collected under a single illumination for training. Firstly, an Illumination Feature Separation Network(IFSNet) is used to remove the illumination components and obtain illumination-invariant features; Then an Absorption Difference Feature Extraction network(ADFENet) is used to reduce the impact of venous structure distortion under illumination conditions, especially overexposure. To replicate the entire range from low-light to overexposure in outdoor scenarios, a novel Multi-Illumination Finger Vein Dataset (MIFVD) is constructed with significant illumination variations. Experimental results show that MDA-Net significantly improves recognition performance under complex illumination conditions, achieving a state-of-the-art (SOTA) average recognition rate of 91.67% and an average equal error rate (EER) of 0.96%. Further validation on public datasets SDU and USM, demonstrates SOTA EERs of 0.16% and 0.10%, respectively. The License for MIFVD can be accessed at: https://github.com/AHU-MedImagingIJR/MIFVD.

Abstract This paper proposes a two‐stage beamforming and combining scheme in frequency division duplex (FDD) massive multiple‐input multiple‐output (MIMO) systems with multi‐antenna users. Specifically, the proposed scheme is a two‐stage scheme, where the pre‐beamforming matrix and pre‐combining matrix are designed using the channel covariance matrix (CCM) in the first stage. The problem of the pre‐beamforming matrix and pre‐combining matrix design are formulated as an 0–1 quadratic integer programming problem. To solve this problem, it is further transformed into an 0–1 mixed linear integer programming problem. In the second stage, the sparse code multiple access and multi‐user precoding and combining are adopted to mitigate the inter‐user interference. Different from the previous transmission scheme using CCM, the proposed scheme consider the multi‐antenna user system, and uses the CCM to design both the pre‐beamforming and pre‐combining matrices to sparsify the effective channel matrix, such that the overhead of pilot and feedback can be reduced. Moreover, the sparse code multiple access can help to better reduce the inter‐user interference. Simulation results validate the good performance of the proposed scheme.

This paper presents a new approach to boost a single-modality (LiDAR) 3D object detector by teaching it to sim-ulate features and responses that follow a multi-modality (LiDAR-image) detector. The approach needs LiDAR-image data only when training the single-modality detector, and once well-trained, it only needs LiDAR data at inference. We design a novel framework to realize the approach: re-sponse distillation to focus on the crucial response samples and avoid most background samples; sparse-voxel distillation to learn voxel semantics and relations from the esti-mated crucial voxels; a fine-grained voxel-to-point distillation to better attend to features of small and distant objects; and instance distillation to further enhance the deep-feature consistency. Experimental results on the nuScenes dataset show that our approach outperforms all SOTA LiDAR-only 3D detectors and even surpasses the baseline LiDAR-image detector on the key NDS metric, filling ~72% mAP gap be-tween the single- and multi-modality detectors.

To boost a detector for single-frame 3D object detection, we present a new approach to train it to simulate features and responses following a detector trained on multi-frame point clouds. Our approach needs multi-frame point clouds only when training the single-frame detector, and once trained, it can detect objects with only single-frame point clouds as inputs during the inference. For this purpose, we design a novel Simulated Multi-Frame Single-Stage object Detector (SMF-SSD) framework: multi-view dense object fusion to densify ground-truth objects to generate a multi-frame point cloud; self-attention voxel distillation to facilitate one-to-many knowledge transfer from multi- to single-frame voxels; multi-scale BEV feature distillation to transfer knowledge in low-level spatial and high-level semantic BEV features; and adaptive response distillation to activate single-frame responses of high confidence and accurate localization. Experimental results on the Waymo test set show that our SMF-SSD consistently outperforms all state-of-the-art single-frame 3D object detectors for all object classes of difficulty levels 1 and 2 in terms of both mAP and mAPH.

Existing single-stage detectors for locating objects in point clouds often treat object localization and category classification as separate tasks, so the localization accuracy and classification confidence may not well align. To address this issue, we present a new single-stage detector named the Confident IoU-Aware Single-Stage object Detector (CIA-SSD). First, we design the lightweight Spatial-Semantic Feature Aggregation module to adaptively fuse high-level abstract semantic features and low-level spatial features for accurate predictions of bounding boxes and classification confidence. Also, the predicted confidence is further rectified with our designed IoU-aware confidence rectification module to make the confidence more consistent with the localization accuracy. Based on the rectified confidence, we further formulate the Distance-variant IoU-weighted NMS to obtain smoother regressions and avoid redundant predictions. We experiment CIA-SSD on 3D car detection in the KITTI test set and show that it attains top performance in terms of the official ranking metric (moderate AP 80.28%) and above 32 FPS inference speed, outperforming all prior single-stage detectors. The code is available at https://github.com/Vegeta2020/CIA-SSD.

We present Self-Ensembling Single-Stage object Detector (SE-SSD) for accurate and efficient 3D object detection in outdoor point clouds. Our key focus is on exploiting both soft and hard targets with our formulated constraints to jointly optimize the model, without introducing extra computation in the inference. Specifically, SE-SSD contains a pair of teacher and student SSDs, in which we design an effective IoU-based matching strategy to filter soft targets from the teacher and formulate a consistency loss to align student predictions with them. Also, to maximize the distilled knowledge for ensembling the teacher, we design a new augmentation scheme to produce shape-aware augmented samples to train the student, aiming to encourage it to infer complete object shapes. Lastly, to better exploit hard targets, we design an ODIoU loss to supervise the student with constraints on the predicted box centers and orientations. Our SE-SSD attains top performance compared with all prior published works. Also, it attains top precisions for car detection in the KITTI benchmark (ranked 1stand 2ndon the BEV and 3D leaderboards1, respectively) with an ultra-high inference speed. The code is available at https://github.com/Vegeta2020/SE-SSD.

With the fast development of effective and low-cost human skeleton capture systems, skeleton-based action recognition has attracted much attention recently. Most existing methods use Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN) to extract spatio-temporal information embedded in the skeleton sequences for action recognition. However, these approaches are limited in the ability of relational modeling in a single skeleton, due to the loss of important structural information when converting the raw skeleton data to adapt to the input format of CNN or RNN. In this paper, we propose an Attentional Recurrent Relational Network-LSTM (ARRN-LSTM) to simultaneously model spatial configurations and temporal dynamics in skeletons for action recognition. We introduce the Recurrent Relational Network to learn the spatial features in a single skeleton, followed by a multi-layer LSTM to learn the temporal features in the skeleton sequences. Between the two modules, we design an adaptive attentional module to focus attention on the most discriminative parts in the single skeleton. To exploit the complementarity from different geometries in the skeleton for sufficient relational modeling, we design a two-stream architecture to learn the structural features among joints and lines simultaneously. Extensive experiments are conducted on several popular skeleton datasets and the results show that the proposed approach achieves better results than most mainstream methods.

Large convolutional networks have achieved impressive classification performances recently. To achieve better performance, convolutional network tends to develop into deeper. However, the increase of network depth causes the linear growth of computational complexity, but cannot bring equivalent increase to the classification accuracy. To alleviate this inconsistence, we propose a cascading approach to accelerate the classification of very deep convolutional neural network. By exploiting the entropy metric to analyze the statistic differences of basic networks between the correctly and mistakenly classified images, we can assign the easily distinguished images to the shallow networks for reducing the computational complexity, and leave the difficultly classified images to the deep networks for maintaining the overall performance. Besides, the proposed cascaded networks can take advantage of the complementarity between different networks, which may boost the classification accuracy compared to the deepest network. We perform the experiments using residual networks of different depths on cifar100 dataset, on the condition of obtaining the similar accuracy to the deepest network, the results show that our cascaded ResNet32-ResNet110 and cascaded ResNet32-ResNet164 can reduce the computation time by 48.6% and 44.3% compared to ResNet110 and ResNet164, respectively. And the cascaded ResNet32-ResNet110-ResNet164 can reduce the computation time by 85.4% compared to the very deep Resnet1001.

Objective: Exploring the regular pattern in combination characteristics of tongue, complexion and pulse in Traditional Chinese Medicine (TCM) and further discussing the standardization of syndrome differentiation of Chronic Gastritis. Methods: Based on 919 Chronic Gastritis cases collected, we combined the algorithms of SM spectral clustering with modularity to identifying sign groups for the involved 112 signs of Chronic Gastritis based on similarity matrix computed by the method of mutual information. Results: The 112 signs were finally clustered into 13 categories. The results of the study suggested that the locations of the disease involved were spleen, stomach, kidney, etc. The general syndrome elements of disease nature included qi asthenia, yang asthenia and blood stasis, etc. Conclusion: These groups consisting of tongue, complexion and pulse identified by spectral clustering with modularity are consistent with clinical experience and TCM theories. It could provide the reference for the research of standardization of the syndrome and syndrome elements differentiation of TCM.

Databases of curated biomedical knowledge, such as the protein-locations reflected in the UniProtKB database, provide an accurate and useful resource to researchers and decision makers. Our goal is to augment the manual efforts currently used to curate knowledge bases with automated approaches that leverage the increased availability of full-text scientific articles. This paper describes experiments that use distant supervised learning to identify protein subcellular localizations, which are important to understand protein function and to identify candidate drug targets. Experiments consider Swiss-Prot, the manually annotated subset of the UniProtKB protein knowledge base, and 43,000 full-text articles from the Journal of Biological Chemistry that contain just under 11.5 million sentences. The system achieves 0.81 precision and 0.49 recall at sentence level and an accuracy of 57% on held-out instances in a test set. Moreover, the approach identifies 8210 instances that are not in the UniProtKB knowledge base. Manual inspection of the 50 most likely relations showed that 41 (82%) were valid. These results have immediate benefit to researchers interested in protein function, and suggest that distant supervision should be explored to complement other manual data curation efforts.

Carrier aggregation (CA) is one technique in Long Term Evolution - Advanced (LTE-A) system, which can extend the bandwidth up to 100MHz and increase the downlink and uplink spectral efficiency. In particular, NxDFT-s-OFDM has been agreed as the uplink bandwidth extension scheme. Nevertheless, the peak-to-average power ratio (PAPR) remains an important issue for user equipment (UE) because it determines the power amplifier (PA) efficiency when multi-carrier transmission is employed. In this paper, a closed-loop component mapping scheme is proposed to solve the PAPR problem caused by the aggregated component carriers (CC). To introduce more diversity gain, a time domain codeword mixing technique is developed to make sure that each Turbo codeword is transmitted through multiple component carriers and multiple antennas. Simulation results show the performance improvements and prove the effective of our proposed approach in terms of PAPR and Block Error Rate (BLER).

In this paper, resource allocation for Orthogonal Frequency Division Multiple Access (OFDMA)-based Wireless Mesh Networks (WMNs) is investigated. First, the Even-Odd framework is modified to be better suited for different traffic patterns. Second, the scheduling problem in the modified Even-Odd framework is formulated as a Mixed Integer Quadratically Constrainted Programming (MIQCP), and a heuristic solution is derived. The proposed heuristic solution is motivated from the fact that the bottleneck collision domain throttles the capacity of the networks, and the links in the bottleneck collision domain should be allocated resource carefully to maximize the throughput. The numerical results show that the proposed scheduling scheme can achieve throughput which is over 90% of the nominal capacity and the modified Even-Odd framework improve the end-to-end throughput for asymmetric traffic.