Beijing Jiaotong University

Gradient Boosting Decision Tree (GBDT) is a popular machine learning algorithm, and has quite a few effective implementations such as XGBoost and pGBRT. Although many engineering optimizations have been adopted in these implementations, the efficiency and scalability are still unsatisfactory when the feature dimension is high and data size is large. A major reason is that for each feature, they need to scan all the data instances to estimate the information gain of all possible split points, which is very time consuming. To tackle this problem, we propose two novel techniques: Gradient-based One-Side Sampling (GOSS) and Exclusive Feature Bundling (EFB). With GOSS, we exclude a significant proportion of data instances with small gradients, and only use the rest to estimate the information gain. We prove that, since the data instances with larger gradients play a more important role in the computation of information gain, GOSS can obtain quite accurate estimation of the information gain with a much smaller data size. With EFB, we bundle mutually exclusive features (i.e., they rarely take nonzero values simultaneously), to reduce the number of features. We prove that finding the optimal bundling of exclusive features is NP-hard, but a greedy algorithm can achieve quite good approximation ratio (and thus can effectively reduce the number of features without hurting the accuracy of split point determination by much). We call our new GBDT implementation with GOSS and EFB LightGBM. Our experiments on multiple public datasets show that, LightGBM speeds up the training process of conventional GBDT by up to over 20 times while achieving almost the same accuracy.

Forecasting the traffic flows is a critical issue for researchers and practitioners in the field of transportation. However, it is very challenging since the traffic flows usually show high nonlinearities and complex patterns. Most existing traffic flow prediction methods, lacking abilities of modeling the dynamic spatial-temporal correlations of traffic data, thus cannot yield satisfactory prediction results. In this paper, we propose a novel attention based spatial-temporal graph convolutional network (ASTGCN) model to solve traffic flow forecasting problem. ASTGCN mainly consists of three independent components to respectively model three temporal properties of traffic flows, i.e., recent, daily-periodic and weekly-periodic dependencies. More specifically, each component contains two major parts: 1) the spatial-temporal attention mechanism to effectively capture the dynamic spatialtemporal correlations in traffic data; 2) the spatial-temporal convolution which simultaneously employs graph convolutions to capture the spatial patterns and common standard convolutions to describe the temporal features. The output of the three components are weighted fused to generate the final prediction results. Experiments on two real-world datasets from the Caltrans Performance Measurement System (PeMS) demonstrate that the proposed ASTGCN model outperforms the state-of-the-art baselines.

The paper presents a novel method, Zero-Reference Deep Curve Estimation (Zero-DCE), which formulates light enhancement as a task of image-specific curve estimation with a deep network. Our method trains a lightweight deep network, DCE-Net, to estimate pixel-wise and high-order curves for dynamic range adjustment of a given image. The curve estimation is specially designed, considering pixel value range, monotonicity, and differentiability. Zero-DCE is appealing in its relaxed assumption on reference images, i.e., it does not require any paired or unpaired data during training. This is achieved through a set of carefully formulated non-reference loss functions, which implicitly measure the enhancement quality and drive the learning of the network. Our method is efficient as image enhancement can be achieved by an intuitive and simple nonlinear curve mapping. Despite its simplicity, we show that it generalizes well to diverse lighting conditions. Extensive experiments on various benchmarks demonstrate the advantages of our method over state-of-the-art methods qualitatively and quantitatively. Furthermore, the potential benefits of our Zero-DCE to face detection in the dark are discussed.

In this paper, we study the salient object detection problem for images. We formulate this problem as a binary labeling task where we separate the salient object from the background. We propose a set of novel features, including multiscale contrast, center-surround histogram, and color spatial distribution, to describe a salient object locally, regionally, and globally. A conditional random field is learned to effectively combine these features for salient object detection. Further, we extend the proposed approach to detect a salient object from sequential images by introducing the dynamic salient features. We collected a large image database containing tens of thousands of carefully labeled images by multiple users and a video segment database, and conducted a set of experiments over them to demonstrate the effectiveness of the proposed approach.

Reconfigurable intelligent surfaces (RISs), also known as intelligent reflecting surfaces (IRSs), or large intelligent surfaces (LISs), <xref ref-type="fn" rid="fn1" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">¹</xref> have received significant attention for their potential to enhance the capacity and coverage of wireless networks by smartly reconfiguring the wireless propagation environment. Therefore, RISs are considered a promising technology for the sixth-generation (6G) of communication networks. In this context, we provide a comprehensive overview of the state-of-the-art on RISs, with focus on their operating principles, performance evaluation, beamforming design and resource management, applications of machine learning to RIS-enhanced wireless networks, as well as the integration of RISs with other emerging technologies. We describe the basic principles of RISs both from physics and communications perspectives, based on which we present performance evaluation of multiantenna assisted RIS systems. In addition, we systematically survey existing designs for RIS-enhanced wireless networks encompassing performance analysis, information theory, and performance optimization perspectives. Furthermore, we survey existing research contributions that apply machine learning for tackling challenges in dynamic scenarios, such as random fluctuations of wireless channels and user mobility in RIS-enhanced wireless networks. Last but not least, we identify major issues and research opportunities associated with the integration of RISs and other emerging technologies for applications to next-generation networks. <fn id="fn1" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><label>¹</label> Without loss of generality, we use the name of RIS in the remainder of this paper. </fn>

Based on empirical survey data, this paper uses concepts from sociotechnical theory and role theory to explore the effects of stress created by information and computer technology (ICT)—that is, "technostress"—on role stress and on individual productivity. We first explain different ways in which ICTs can create stress in users and identify factors that create technostress. We next propose three hypotheses: (1) technostress is inversely related to individual productivity, (2) role stress is inversely related to individual productivity, and (3) technostress is directly related to role stress. We then use structural equation modeling on survey data from ICT users in 223 organizations to test the hypotheses. The results show support for them. Theoretically, the paper contributes in three ways. First, the different dimensions of technostress identified here add to existing concepts on stress experienced by individuals in organizations. Second, by showing that technostress inversely affects productivity, the paper reinforces that failure to manage the effects of ICT-induced stress can offset expected increases in productivity. Third, validation of the positive relationship between technostress and role stress adds a new conceptual thread to literature analyzing the relationship between technology and organizational roles and structure. In the practical domain, the paper proposes a diagnostic tool to evaluate the extent to which technostress is present in an organization and suggests that the adverse effects of technostress can be partly countered by strategies that reduce role conflict and role overload.

Green-luminescent functional graphene quantum dots (GQDs) are prepared by a facile electrochemical approach. The GQDs are rich in oxygen-containing functional groups and soluble in aqueous or organic media, facilitating further functionalization and various applications. Polymer photovoltaic devices using GQDs as a new type of electron-acceptor material are also demonstrated. Detailed facts of importance to specialist readers are published as ”Supporting Information”. Such documents are peer-reviewed, but not copy-edited or typeset. They are made available as submitted by the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Spatial-temporal network data forecasting is of great importance in a huge amount of applications for traffic management and urban planning. However, the underlying complex spatial-temporal correlations and heterogeneities make this problem challenging. Existing methods usually use separate components to capture spatial and temporal correlations and ignore the heterogeneities in spatial-temporal data. In this paper, we propose a novel model, named Spatial-Temporal Synchronous Graph Convolutional Networks (STSGCN), for spatial-temporal network data forecasting. The model is able to effectively capture the complex localized spatial-temporal correlations through an elaborately designed spatial-temporal synchronous modeling mechanism. Meanwhile, multiple modules for different time periods are designed in the model to effectively capture the heterogeneities in localized spatial-temporal graphs. Extensive experiments are conducted on four real-world datasets, which demonstrates that our method achieves the state-of-the-art performance and consistently outperforms other baselines.

This paper proposes a convolutional neural network (CNN)-based method that learns traffic as images and predicts large-scale, network-wide traffic speed with a high accuracy. Spatiotemporal traffic dynamics are converted to images describing the time and space relations of traffic flow via a two-dimensional time-space matrix. A CNN is applied to the image following two consecutive steps: abstract traffic feature extraction and network-wide traffic speed prediction. The effectiveness of the proposed method is evaluated by taking two real-world transportation networks, the second ring road and north-east transportation network in Beijing, as examples, and comparing the method with four prevailing algorithms, namely, ordinary least squares, k-nearest neighbors, artificial neural network, and random forest, and three deep learning architectures, namely, stacked autoencoder, recurrent neural network, and long-short-term memory network. The results show that the proposed method outperforms other algorithms by an average accuracy improvement of 42.91% within an acceptable execution time. The CNN can train the model in a reasonable time and, thus, is suitable for large-scale transportation networks.

This paper gives an overview of medium-voltage (MV) multilevel converters with a focus on achieving minimum harmonic distortion and high efficiency at low switching frequency operation. Increasing the power rating by minimizing switching frequency while still maintaining reasonable power quality is an important requirement and a persistent challenge for the industry. Existing solutions are discussed and analyzed based on their topologies, limitations, and control techniques. As a preferred option for future research and application, an inverter configuration based on three-level building blocks to generate five-level voltage waveforms is suggested. This paper shows that such an inverter may be operated at a very low switching frequency to achieve minimum on-state and dynamic device losses for highly efficient MV drive applications while maintaining low harmonic distortion.

Functionalization is an important way to breed new properties and applications for a material. This review presents an overview of the progresses in functionalized hexagonal boron nitride (h-BN) nanomaterials. It begins with an introduction of h-BN structural features, physical and chemical properties, followed by an emphasis on the developments of BN functionalization strategies and its emerging properties/applications, and ends with the research perspectives. Different functionalization methods, including physical and chemical routes, are comprehensively described toward fabrication of various BN derivatives, hetero- and porous structures, etc. Novel properties of functionalized BN materials, such as high water solubility, excellent biocompatibility, tunable surface affinities, good processibility, adjustable band gaps, etc., have guaranteed wide applications in biomedical, electronic, composite, environmental and "green" energy-related fields.

Big data is becoming a research focus in intelligent transportation systems (ITS), which can be seen in many projects around the world. Intelligent transportation systems will produce a large amount of data. The produced big data will have profound impacts on the design and application of intelligent transportation systems, which makes ITS safer, more efficient, and profitable. Studying big data analytics in ITS is a flourishing field. This paper first reviews the history and characteristics of big data and intelligent transportation systems. The framework of conducting big data analytics in ITS is discussed next, where the data source and collection methods, data analytics methods and platforms, and big data analytics application categories are summarized. Several case studies of big data analytics applications in intelligent transportation systems, including road traffic accidents analysis, road traffic flow prediction, public transportation service plan, personal travel route plan, rail transportation management and control, and assets maintenance are introduced. Finally, this paper discusses some open challenges of using big data analytics in ITS.

Underwater images suffer from color casts and low contrast due to wavelength- and distance-dependent attenuation and scattering. To solve these two degradation issues, we present an underwater image enhancement network via medium transmission-guided multi-color space embedding, called Ucolor. Concretely, we first propose a multi-color space encoder network, which enriches the diversity of feature representations by incorporating the characteristics of different color spaces into a unified structure. Coupled with an attention mechanism, the most discriminative features extracted from multiple color spaces are adaptively integrated and highlighted. Inspired by underwater imaging physical models, we design a medium transmission (indicating the percentage of the scene radiance reaching the camera)-guided decoder network to enhance the response of network towards quality-degraded regions. As a result, our network can effectively improve the visual quality of underwater images by exploiting multiple color spaces embedding and the advantages of both physical model-based and learning-based methods. Extensive experiments demonstrate that our Ucolor achieves superior performance against state-of-the-art methods in terms of both visual quality and quantitative metrics. The code is publicly available at: https://li-chongyi.github.io/Proj_Ucolor.html.

We investigate a principle way to progressively mine discriminative object regions using classification networks to address the weakly-supervised semantic segmentation problems. Classification networks are only responsive to small and sparse discriminative regions from the object of interest, which deviates from the requirement of the segmentation task that needs to localize dense, interior and integral regions for pixel-wise inference. To mitigate this gap, we propose a new adversarial erasing approach for localizing and expanding object regions progressively. Starting with a single small object region, our proposed approach drives the classification network to sequentially discover new and complement object regions by erasing the current mined regions in an adversarial manner. These localized regions eventually constitute a dense and complete object region for learning semantic segmentation. To further enhance the quality of the discovered regions by adversarial erasing, an online prohibitive segmentation learning approach is developed to collaborate with adversarial erasing by providing auxiliary segmentation supervision modulated by the more reliable classification scores. Despite its apparent simplicity, the proposed approach achieves 55.0% and 55.7% mean Intersection-over-Union (mIoU) scores on PASCAL VOC 2012 val and test sets, which are the new state-of-the-arts.

Multiple antenna technologies have attracted much research interest for several decades and have gradually made their way into mainstream communication systems. Two main benefits are adaptive beamforming gains and spatial multiplexing, leading to high data rates per user and per cell, especially when large antenna arrays are adopted. Since multiple antenna technology has become a key component of the fifth-generation (5G) networks, it is time for the research community to look for new multiple antenna technologies to meet the immensely higher data rate, reliability, and traffic demands in the beyond 5G era. Radically new approaches are required to achieve orders-of-magnitude improvements in these metrics. There will be large technical challenges, many of which are yet to be identified. In this paper, we survey three new multiple antenna technologies that can play key roles in beyond 5G networks: cell-free massive MIMO, beamspace massive MIMO, and intelligent reflecting surfaces. For each of these technologies, we present the fundamental motivation, key characteristics, recent technical progresses, and provide our perspectives for future research directions. The paper is not meant to be a survey/tutorial of a mature subject, but rather serve as a catalyst to encourage more research and experiments in these multiple antenna technologies.

Internet of Things (IoT) computing offloading is a challenging issue, especially in remote areas where common edge/cloud infrastructure is unavailable. In this paper, we present a space-air-ground integrated network (SAGIN) edge/cloud computing architecture for offloading the computation-intensive applications considering remote energy and computation constraints, where flying unmanned aerial vehicles (UAVs) provide near-user edge computing and satellites provide access to the cloud computing. First, for UAV edge servers, we propose a joint resource allocation and task scheduling approach to efficiently allocate the computing resources to virtual machines (VMs) and schedule the offloaded tasks. Second, we investigate the computing offloading problem in SAGIN and propose a learning-based approach to learn the optimal offloading policy from the dynamic SAGIN environments. Specifically, we formulate the offloading decision making as a Markov decision process where the system state considers the network dynamics. To cope with the system dynamics and complexity, we propose a deep reinforcement learning-based computing offloading approach to learn the optimal offloading policy on-the-fly, where we adopt the policy gradient method to handle the large action space and actor-critic method to accelerate the learning process. Simulation results show that the proposed edge VM allocation and task scheduling approach can achieve near-optimal performance with very low complexity and the proposed learning-based computing offloading algorithm not only converges fast but also achieves a lower total cost compared with other offloading approaches.

The last eight years (2009-2017) have seen an explosive growth of interest in organic-inorganic halide perovskites in the research communities of photovoltaics and light-emitting diodes. In addition, recent advancements have demonstrated that this type of perovskite has a great potential in the technology of light-signal detection with a comparable performance to commercially available crystalline Si and III-V photodetectors. The contemporary growth of state-of-the-art multifunctional perovskites in the field of light-signal detection has benefited from its outstanding intrinsic optoelectronic properties, including photoinduced polarization, high drift mobilities, and effective charge collection, which are excellent for this application. Photoactive perovskite semiconductors combine effective light absorption, allowing detection of a wide range of electromagnetic waves from ultraviolet and visible, to the near-infrared region, with low-cost solution processability and good photon yield. This class of semiconductor might empower breakthrough photodetector technology in the field of imaging, optical communications, and biomedical sensing. Therefore, here, the focus is specifically on the critical understanding of materials synthesis, design, and engineering for the next-stage development of perovskite photodetectors and highlighting the current challenges in the field, which need to be further studied in the future.

Convolutional Neural Network (CNN) has demonstrated promising performance in single-label image classification tasks. However, how CNN best copes with multi-label images still remains an open problem, mainly due to the complex underlying object layouts and insufficient multi-label training images. In this work, we propose a flexible deep CNN infrastructure, called Hypotheses-CNN-Pooling (HCP), where an arbitrary number of object segment hypotheses are taken as the inputs, then a shared CNN is connected with each hypothesis, and finally the CNN output results from different hypotheses are aggregated with max pooling to produce the ultimate multi-label predictions. Some unique characteristics of this flexible deep CNN infrastructure include: 1) no ground-truth bounding box information is required for training; 2) the whole HCP infrastructure is robust to possibly noisy and/or redundant hypotheses; 3) the shared CNN is flexible and can be well pre-trained with a large-scale single-label image dataset, e.g., ImageNet; and 4) it may naturally output multi-label prediction results. Experimental results on Pascal VOC 2007 and VOC 2012 multi-label image datasets well demonstrate the superiority of the proposed HCP infrastructure over other state-of-the-arts. In particular, the mAP reaches 90.5% by HCP only and 93.2% after the fusion with our complementary result in [44] based on hand-crafted features on the VOC 2012 dataset.

Computation offloading services provide required computing resources for vehicles with computation-intensive tasks. Past computation offloading research mainly focused on mobile edge computing (MEC) or cloud computing, separately. This paper presents a collaborative approach based on MEC and cloud computing that offloads services to automobiles in vehicular networks. A cloud-MEC collaborative computation offloading problem is formulated through jointly optimizing computation offloading decision and computation resource allocation. Since the problem is non-convex and NP-hard, we propose a collaborative computation offloading and resource allocation optimization (CCORAO) scheme, and design a distributed computation offloading and resource allocation algorithm for CCORAO scheme that achieves the optimal solution. The simulation results show that the proposed algorithm can effectively improve the system utility and computation time, especially for the scenario where the MEC servers fail to meet demands due to insufficient computation resources.

The class activation maps are generated from the final convolutional layer of CNN. They can highlight discriminative object regions for the class of interest. These discovered object regions have been widely used for weakly-supervised tasks. However, due to the small spatial resolution of the final convolutional layer, such class activation maps often locate coarse regions of the target objects, limiting the performance of weakly-supervised tasks that need pixel-accurate object locations. Thus, we aim to generate more fine-grained object localization information from the class activation maps to locate the target objects more accurately. In this paper, by rethinking the relationships between the feature maps and their corresponding gradients, we propose a simple yet effective method, called LayerCAM. It can produce reliable class activation maps for different layers of CNN. This property enables us to collect object localization information from coarse (rough spatial localization) to fine (precise fine-grained details) levels. We further integrate them into a high-quality class activation map, where the object-related pixels can be better highlighted. To evaluate the quality of the class activation maps produced by LayerCAM, we apply them to weakly-supervised object localization and semantic segmentation. Experiments demonstrate that the class activation maps generated by our method are more effective and reliable than those by the existing attention methods. The code will be made publicly available.