The smart vehicles construct Internet of Vehicle (IoV), which can execute various intelligent services. Although the computation capability a vehicle is limited, multi-type edge computing nodes provide heterogeneous resources for vehicular When offloading complex service to node, decision its destination should be considered according numerous factors. This paper mostly formulate as resource scheduling problem with single or multiple objective function and constraints, where some customized...

Cross-domain sentiment classification aims to address the lack of massive amounts labeled data. It demands predict polarity on a target domain utilizing classifier learned from source domain. In this paper, we investigate how efficiently apply pre-training language model BERT unsupervised adaptation. Due task and corpus, is task-agnostic, which lacks awareness can not distinguish characteristic when transferring knowledge. To tackle these problems, design post-training procedure, contains...

We report 55-nm gate AlInN/GaN high-electron-mobility transistors (HEMTs) featuring a short-circuit current gain cutoff frequency of fT = 205 GHz at room temperature, new record for GaN-based HEMTs. The devices source maximum density 2.3 A/mm V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">GS</sub> 0 and show measured transconductance 575 mS/mm, which is the highest value reported to date nonrecessed nitride Comparison state-of-the-art...

Network telemetry is essential for administrators to monitor massive data traffic in a network-wide manner. Existing solutions often face the dilemma between resource efficiency (i.e., low CPU, memory, and bandwidth overhead) full accuracy error-free holistic measurement). We break this via architectural design OmniMon, which simultaneously achieves flow-level large-scale centers. OmniMon carefully coordinates collaboration among different types of entities whole network execute operations,...

Federated learning is an emerging concept that trains the machine models with local distributed data sets, without sending raw to center. But, in Internet of Things (IoT) where wireless network resource constrained, key problem federated communication overhead for parameter synchronization, which wastes bandwidth, increases training time, and even impacts model accuracy. Gradient sparsification has received increasing attention, only updates significant gradients accumulates insignificant...

Solvothermal reaction of cadmium nitrate with a polypyridine-containing ligand tetra(pyridine-4-yl)benzene-1,4-diamine (TBD) gives rise to novel Cd(II)-organic framework the 3D interpenetrated architecture [Cd(TBD) (NO3)2·DMA](1) (DMA = N,N-dimethylacetamide). The stability test is performed on compound 1, which shows that Cd-MOF can be quite stable in water and even harsh conditions. A photoluminescence study has been investigated 1. Remarkably, nitrofurantoin, as kind antibiotic, great...

Chunning Du, Haifeng Sun, Jingyu Wang, Qi Qi, Jianxin Liao, Tong Xu, Ming Liu. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and 9th International Joint (EMNLP-IJCNLP). 2019.

In this paper, we propose an adaptive weighting regression (AWR) method to leverage the advantages of both detection-based and regression-based method. Hand joint coordinates are estimated as discrete integration all pixels in dense representation, guided by weight maps. This learnable aggregation process introduces supervision that allows end-to-end training brings adaptability maps, making network more accurate robust. Comprehensive exploration experiments conducted validate effectiveness...

Channel pruning can significantly accelerate and compress deep neural networks. Many channel works utilize structured sparsity regularization to zero out all the weights in some channels automatically obtain structure-sparse network training stage. However, these methods apply on each layer separately where correlations between consecutive layers are omitted. In this paper, we first combine one out-channel current corresponding in-channel next as a group, namely out-in-channel. Our proposed...

The Internet of Vehicles (IoV) as a promising application Things (IoT) has played significant role in autonomous driving, by connecting intelligent vehicles. Autonomous driving needs to process the mass environmental sensing data coordination with surrounding vehicles, and makes an accurate judgment accordingly. Since vehicles always have limited computing resources, processing these parallel efficient task scheduling is one most important topics. Most current work focuses on formulating...

In 6G mobile systems, network slicing is an emerging technology to support services with distinct requirements by dividing a common infrastructure into multiple logical networks. However, as management method, it difficult for achieve real-time resource allocation satisfy the stringent requirement of in This paper introduces joint and routing mechanism, which combines control framework provide fine-grained, dynamic allocation. Graph Convolutional Networks (GCN)-powered Multi-Task Deep...

We report the fabrication and characterization of 30-nm-gate fully passivated AlInN/GaN high-electron mobility transistors (HEMTs) with cutoff frequencies <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">f</i> <sub xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> xmlns:xlink="http://www.w3.org/1999/xlink">MAX</sub> simultaneously exceeding 200 GHz at a given bias point. The current gain frequency does not vary significantly for 2.5 <;...

Federated learning is an emerging concept that trains the machine models with distributed datasets, without sending raw data to center. But in edge computing enviroment where wireless network resource constrained, key problem of federated communication overhead for parameters synchronization, which wastes bandwidth, increases training time, and even impacts model accuracy. Gradient sparsification has received increasing attention, only updates significant gradients accumulates insignificant...

With the development of Internet Things (IoT) and 5G, there are ubiquitous smart devices network functions providing emerging services efficiently optimally through building many connections based on WiFi, LTE/5G, Ethernet, etc. The Multipath TCP (MPTCP) protocol that enables these to establish multiple paths for simultaneous data transmission, has been a widely used extension standard in functions. On other hand, more heavy time-varying traffic loads generated an MPTCP network, so efficient...

Depth images and point clouds are the two most commonly used data representations for depth-based 3D hand pose estimation. Benefiting from structuring of image inherent inductive biases 2D Convolutional Neural Network (CNN), image-based methods highly efficient effective. However, treating depth as a inevitably ignores nature data. Point cloud-based can better mine geometric structure these suffer disorder non-structure cloud data, which is computationally inefficient. In this paper, we...

Grown on a (111) high-resistivity silicon substrate, 0.1-mum gate AlInN/GaN high-electron mobility transistors (HEMTs) achieve maximum current density of 1.3 A/mm, an extrinsic transconductance 330 mS/mm, and peak gain cutoff frequency as high f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> = 102 GHz, which is the highest value reported so far for nitride-based devices substrates, well any AlInN/GaN-based HEMT regardless substrate...

We report high-speed fully passivated deep submicrometer (Al,Ga)N/GaN high-electron mobility transistors (HEMTs) grown on (111) high-resistivity silicon with current gain cutoff frequencies of as high f <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sub> = 107 GHz and maximum oscillation reaching xmlns:xlink="http://www.w3.org/1999/xlink">MAX</sub> 150 GHz. Together, these are the highest values achieved for GaN-based HEMTs implemented...

Massive logs contain crucial information about the working status of software systems, which contributes to anomaly detection and troubleshooting. For engineers, it is a laborious task manually inspect raw know system running status, therefore an automated log summarization tool can be helpful. However, due specificity in terms grammar, vocabulary semantics, existing natural language-based methods cannot perform well analysis. To address these issues, we propose LogNotion, general framework...

Temporal sentence grounding in videos (TSGV) faces challenges due to public TSGV datasets containing significant temporal biases, which are attributed the uneven distributions of target moments. Existing methods generate augmented videos, where moments forced have varying locations. However, since video lengths given small variations, only changing locations results poor generalization ability with lengths. In this paper, we propose a novel training framework complemented by diversified data...