Recent research has demonstrated the potential of deploying deep neural networks (DNNs) on resource-constrained mobile platforms by trimming down network complexity using different compression techniques. The current practice only investigate stand-alone schemes even though each technique may be well suited for certain types DNN layers. Also, these techniques are optimized merely inference accuracy DNNs, without explicitly considering other application-driven system performance (e.g. latency...

Underwater target detection is an indispensable part of marine environmental engineering and a fast accurate method detecting underwater targets essential. Although many algorithms have achieved great accuracy in daily scenes, there are issues low-quality images due to the complex environment, which makes applying these deep learning directly process tasks difficult. In this paper, we presented algorithm for based on improved You Only Look Once (YOLO) v4 response environment. First,...

In this article, we focus on solving the energy optimization problem for real-time streaming applications multiprocessor System-on-Chip by combining task-level coarse-grained software pipelining with DVS (Dynamic Voltage Scaling) and DPM Power Management) considering transition overhead, inter-core communication discrete voltage levels. We propose a two-phase approach to solve problem. first phase, task parallelization algorithm called RDAG transform periodic dependent graph into set of...

Radar has long been a common sensor on autonomous vehicles for obstacle ranging and speed estimation. However, as robust to all-weather conditions, radar's capability not well-exploited, compared with camera or LiDAR. Instead of just serving supplementary sensor, rich information hidden in the radio frequencies can potentially provide useful clues achieve more complicated tasks, like object classification detection. In this paper, we propose new dataset, named CRUW <sup...

Distracted driving behavior is known as a leading factor in road traffic injuries and deaths. Fortunately, rapidly developing deep learning technology has shown its potential distracted detection. Nevertheless, learning-based solutions need to collect large amounts of data captured by camera sensors the vehicle, which will cause serious privacy concerns. As privacy-preserving distributed paradigm, federated (FL) achieved competitive performance many applications recently. Inspired this, we...

Mobile edge computing introduces a novel paradigm for mobile devices, reducing execution latency and energy consumption by offloading tasks to servers or other idle devices. In this paper, we consider the utility optimization problem of two typical tasks, latency-sensitive latency-tolerant among multiple devices base stations. can choose three modes optimize utility: local computing, task allocation stations, through device-to-device communication. To address problem, formalize it as...

Recent breakthroughs in deep neural networks (DNNs) have fueled a tremendously growing demand for bringing DNN-powered intelligence into mobile platforms. While the potential of deploying DNNs on resource-constrained platforms has been demonstrated by DNN compression techniques, current practice suffers from two limitations: 1) merely stand-alone schemes are investigated even though each technique only suit certain types layers; and 2) mostly techniques optimized DNNs' inference accuracy,...

Unmanned aerial vehicles (UAVs) are beginning to make a splash in emergency disaster scenarios owing its excellent air mobility and flexibility. Considering that large base stations often cannot be deployed areas the first place variation of communication links between UAVs, we formulate task scheduling problem for as two-stage Lyapunov optimization propose dispersed computing network consisting UAVs ground mobile devices, which is used collaborative computing. We decouple long-term...

In this paper, we focus on joint energy and performance optimization for streaming applications multiprocessor systems-on-chip by combining task-level coarse-grained software pipelining with DVS (dynamic voltage scaling)and DPM power management) techniques the considerations of transition overhead, inter-processor communication discrete levels. We propose a two-phase approach to solve problem. first phase, task parallelization algorithm called RDAG transform periodic dependent graph into set...

Message authentication is one of the most effective ways to thwart unauthorized messages from being transmitted in wireless ad-hoc networks such as sensor (WSNs) and vehicular (VANETs). For this reason, many message schemes have been developed, based on either symmetric-key cryptosystems or public-key cryptosystems. Public-key simpler key management are easier scale, thus more suitable for authentication. Among all kinds cryptosystems, identity cryptosystem attractive since public keys can...

The goal of relational triple extraction is to extract knowledge-rich triples from unstructured text. Although the previous methods obtain considerable performance, there are still some problems, such as error propagation, overlapping problem, and suboptimal subject–object alignment. To address shortcomings above, in this paper, we decompose task into three subtasks a fresh perspective: entity extraction, alignment relation judgement, well propose novel bi-directional translating decoding...

In the real world, speaker recognition systems usually suffer from serious performance degradation due to domain mismatch between training and test conditions. To alleviate harmful effect of shift, unsupervised adaptation methods are introduced learn domain-invariant representations, which focus on addressing single-source-to-single-target issue. However, labeled data collected multiple sources, such as different languages, genres devices. The single-domain can not deal with complex...

This paper proposes a 3D vehicle-detection algorithm based on multimodal feature fusion to address the problem of low accuracy in unmanned system environment awareness. The matches coordinate relationships between two sensors and reduces sampling errors by combining millimeter-wave radar camera calibration. Statistical filtering is used remove redundant points from data reduce outlier interference; module constructed fuse point cloud image information using pixel-by-pixel averaging....

In this paper, we focus on designing and developing ProMETheus, an intelligent system for meeting minutes generated from audio data. The first task in ProMETheus is to recognize the speakers noisy Speaker recognition algorithm used automatically identify who speaking according speech Naturally, will transcribe speakers' text so that can generate complete with name chronologically. order show subject of agreed action, use summarization extract meaningful key phrases summary sentences text....