Within the context of cleaner production, enhancing energy efficiency and sustainability has emerged as a central focus in supercomputing center development. To address the challenges in predicting energy consumption,...
详细信息
Objective and effective algorithm performance evaluation results are an important basis for the selection of tracking algorithms. Problems in the existing performance evaluation of moving target tracking algorithms in...
详细信息
As a kind of keywords to describe video contents, tags are extremely beneficial for viewers to locate videos when they search the site. Additionally, tags also help video platform operators to better organize and reco...
详细信息
ISBN:
(纸本)9781665417631
As a kind of keywords to describe video contents, tags are extremely beneficial for viewers to locate videos when they search the site. Additionally, tags also help video platform operators to better organize and recommend videos to platform users. Previous approaches are mainly based on manually tagging or tag propagation via video content analysis, which is time-consuming and resource-consuming. Especially, given the high volume of fresh movies generated per year, it is difficult to accurately tag all the movies manually. Moreover, video content analysis is also not easy considering typical features (e.g., long duration, complicated scenarios) of commercial movies. In this paper, we propose an automatic tagging algorithm called TagRec that exploits crowdsourced user reviews to generate accurate movie tags. We observe that user reviews contain rich information about movies (e.g., quality, actors) which can be learned to generate high-quality movie tags. Inspired by the above observation, we choose to transform the movie video tagging problem into a tag recommendation problem, in which tags are recommended to different movies by extracting knowledge from crowdsourced movie reviews. We take latent topics, tag co-occurrence probability and tag semantics into account, and formulate the problem as a recommendation optimization problem. We evaluate the performance of our pro-posed TagRec algorithm with a large-scale real-world dataset. Extensive experiments demonstrate that TagRec achieves 7.1 % and 9.6% improvement compared with other state-of-the-art methods in terms of Hit Ratio and Normalized Discounted Cumulative Gain respectively.
The algal bloom is a prominent manifestation of water pollution. Synthetic aperture radar (SAR) shows an advantage in water monitoring due to its characteristic of all-time and all-weather. The water regions where alg...
详细信息
Unmanned aerial vehicles (UAVs) can be used as air base stations to provide fast wireless connections for ground users. Due to their constraints on both mobility and energy consumption, a key problem is how to deploy ...
详细信息
ISBN:
(数字)9781728194844
ISBN:
(纸本)9781728194851
Unmanned aerial vehicles (UAVs) can be used as air base stations to provide fast wireless connections for ground users. Due to their constraints on both mobility and energy consumption, a key problem is how to deploy UAVs adaptively in a geographic area with changing traffic demand of mobile users, while meeting the aforemetioned constraints. In this paper, we propose an adaptive deployment strategy for UAV-aided networks based on hybrid deep reinforcement learning, where a UAV can adjust its movement direction and distance to serve users who move randomly in the target area. Through hybrid deep reinforcement learning, UAVs can be trained offline to obtain the global state information and learn a completely distributed control strategy, with which each UAV only needs to take actions based on its observed state in the real deployment to be fully adaptive. Moreover, in order to improve the speed and effect of learning, we improve hybrid reinforcement learning, by adding genetic algorithms and TD-error-based resampling optimization mechanism. Simulation results show that the hybrid deep reinforcement learning algorithm has better efficiency and robustness in multi-UAV control, and has better performance in terms of coverage, energy consumption and average throughput, by which average throughput can be increased by 20% to 60%.
Objectives: To investigate whether the pleurae, airways and vessels surrounding a nodule on non-contrast computed tomography (CT) can discriminate benign and malignant pulmonary nodules. Materials and Methods: The LID...
详细信息
Multispectral pan-sharpening aims at producing a high resolution (HR) multispectral (MS) image in both spatial and spectral domains by fusing a panchromatic (PAN) image and a corresponding MS image. In this paper, we ...
详细信息
H 2 -free semi-hydrogenation at room temperature shows great advantage for replacing the thermocatalytic process in industry owing to the high energy and resource saving, however, remains great challenges. Herein, a t...
详细信息
H 2 -free semi-hydrogenation at room temperature shows great advantage for replacing the thermocatalytic process in industry owing to the high energy and resource saving, however, remains great challenges. Herein, a tree-like Pd dendrites array decorated Pd membrane was constructed as the core device in an electrochemistry assisted gas-fed membrane reactor for butadiene semi-hydrogenation. It reveals that hydrogen atomic sieving effect of this Pd-based membrane under electrochemical condition was the key for semi-hydrogenation. The configuration study of Pd nanostructured membrane demonstrates that the penetration of hydrogen atoms through Pd membrane from electrochemical side to chemical side is affected by the consumption of hydrogen atom in semi-hydrogenation step. Such atomic sieving property of nanostructured Pd membrane with 5.1 times increase in catalytic active surface area brings above 14 times higher in butadiene conversion than that of bare Pd foil, with ≈90 % of butenes selectivity at butadiene conversion ≈98 % over 300 h of H 2 -free reaction under 15 mA cm −2 .
Cooperative coupling of H 2 evolution with oxidative organic synthesis is promising in avoiding the use of sacrificial agents and producing hydrogen energy with value-added chemicals simultaneously. Nonetheless, the p...
详细信息
Cooperative coupling of H 2 evolution with oxidative organic synthesis is promising in avoiding the use of sacrificial agents and producing hydrogen energy with value-added chemicals simultaneously. Nonetheless, the photocatalytic activity is obstructed by sluggish electron-hole separation and limited redox potentials. Herein, Ni-doped Zn 0.2 Cd 0.8 S quantum dots are chosen after screening by DFT simulation to couple with TiO 2 microspheres, forming a step-scheme heterojunction. The Ni-doped configuration tunes the highly active S site for augmented H 2 evolution, and the interfacial Ni−O bonds provide fast channels at the atomic level to lower the energy barrier for charge transfer. Also, DFT calculations reveal an enhanced built-in electric field in the heterojunction for superior charge migration and separation. Kinetic analysis by femtosecond transient absorption spectra demonstrates that expedited charge migration with electrons first transfer to Ni 2+ and then to S sites. Therefore, the designed catalyst delivers drastically elevated H 2 yield (4.55 mmol g −1 h −1 ) and N-benzylidenebenzylamine production rate (3.35 mmol g −1 h −1 ). This work provides atomic-scale insights into the coordinated modulation of active sites and built-in electric fields in step-scheme heterojunction for ameliorative photocatalytic performance.
Depth modal features can provide complementary information for salient object detection (SOD). Most of the existing RGB-D SOD methods focus on fully combining RGB and Depth modal features without distinguishing them. ...
Depth modal features can provide complementary information for salient object detection (SOD). Most of the existing RGB-D SOD methods focus on fully combining RGB and Depth modal features without distinguishing them. In this paper, we propose a new depth guided cross-modal residual adaptive network for RGB-D SOD. We use two independent resnet-50 to extract the features of the two modes respectively. Then the cross-modal channel-wise refinement module is designed to obtain complementary modal information. We design a crossmodal guided module to make complementary modal information guide RGB image feature extraction. Finally, the residual adaptive selection module is used to enhance the spatial mutual concerns between the two modal features to achieve multimodal information fusion. Experimental results show that our method can achieve a more reasonable fusion state of RGB and Depth, and verify the effectiveness of our final saliency model.
暂无评论