The explosion of Next Generation Sequencing (NGS) data with over one billion reads per day poses a great challenge to the capability current computing systems. In this paper, we proposed CPU-FPGA heterogeneous architecture for accelerating short mapping algorithm, which was built upon concept hash-index. particular, by extracting and most time-consuming basic operations specialized processing elements (PEs), our new algorithm is favorable efficient acceleration on FPGAs. implemented...

In fisheye images, rich distinct distortion patterns are regularly distributed in the image plane. These independent of visual content and provide informative cues for rectification. To make best such rectification cues, we introduce SimFIR, a simple framework based on self-supervised representation learning. Technically, first split into multiple patches extract their representations with Vision Transformer (ViT). learn fine-grained representations, then associate different specific model,...

Over the past decades, we noticed huge advances in FPGA technologies. The topic of floating-point accelerator on has gained renewed interests due to increased device size and emergence fast hardware library. popularity FFT makes it easier justify spending lots effort doing detailed optimization. However, ever increasing data some compelling application domains remains beyond capability existing accelerators. demand for more performance an active research topic. In this paper, leveraging...

MmWave communication is a promising technique of future 5G system. Large bandwidth and high directional gain are the two advantages mmWave solutions. High leads usage massive units. The system maintenance status management play more important role in wireless systems contribute higher operation cost than before. In our paper we proposed fault finding location methods based on deep neural network (DNN) with different complexity, simple designed to low cost, other precision start when former...

Camera lenses often suffer from optical aberrations, causing radial distortion in the captured images. In those images, there exists a clear and general physical model. However, existing solutions, such rich geometric prior is under-utilized, formulation of an effective prediction target under-explored. To this end, we introduce Radial Distortion TRansformer (RDTR), new framework for rectification. Our RDTR includes model-aware pre-training stage feature extraction deformation estimation...

We present assertion based question answering (ABQA), an open domain task that takes a and passage as inputs, outputs semi-structured consisting of subject, predicate list arguments. An conveys more evidences than short answer span in reading comprehension, it is concise tedious passage-based QA. These advantages make ABQA suitable for human-computer interaction scenarios such voice-controlled speakers. Further progress towards improving requires richer supervised dataset powerful models...

Abstract New services, such as distributed photovoltaic regulation and control, pose new service requirements for communication networks in the power system. These include low latency, high reliability, large bandwidth. Consequently, heterogeneous face challenge of maintaining quality (QoS) while enhancing network resource utilization. Therefore, this paper puts forward a highly efficient optimization algorithm slicing scheduling networks. Our first step involves establishing an...

The emerging next-generation sequencing method based on PCR technology boosts genome speed considerably, the expense is also get decreased. It has been utilized to address a broad range of bioinformatics problems. Limited by reliable output sequence length technologies, we are confined study gene fragments with 30 - 50 bps in general and it relatively shorter than traditional fragment length. Anchoring long reference an essential prerequisite step for further assembly analysis works. Due...

The emergence of embedded and multimedia applications, which have a data-centric favor to them, great influences on the design methodology future systems. 2D FFT is particular importance these applications. In this paper, leveraging reconfigurable features off-the-shelf FPGAs, we propose stream architecture that suitable for accelerating with variable non-power-of-two problem size. system built upon concept coarse-grained application-specific pipeline optimized real data specifically....

The wide acceptance of bioinformatics, medical imaging and multimedia applications, which have a data-centric favor to them, require more efficient application-specific systems be built. Due the advances in modern FPGA technologies recently, there has been resurgence research aimed at accelerator design that leverages FPGAs accelerate large-scale scientific applications. In this paper, we exploit trend towards FPGA-based provide proof-of-concept comprehensive case study on for...

We present assertion based question answering (ABQA), an open domain task that takes a and passage as inputs, outputs semi-structured consisting of subject, predicate list arguments. An conveys more evidences than short answer span in reading comprehension, it is concise tedious passage-based QA. These advantages make ABQA suitable for human-computer interaction scenarios such voice-controlled speakers. Further progress towards improving requires richer supervised dataset powerful models...

The data deluge in medical imaging processing requires faster and more efficient systems. Due to the advance recent heterogeneous architecture, there has been a resurgence research aimed at domain-specific accelerators. In this article, we develop an experimental system SuperDragon for evaluating acceleration of single-particle Cryo-electron microscopy (Cryo-EM) 3D reconstruction package EMAN through hybrid CPU, GPU, FPGA parallel architecture. Based on comprehensive workload...

It is undeniable that the concepts of artificial intelligence and blockchain are spreading at an alarming speed. These two technologies have different degrees technical complexity multi-dimensional business meaning. The convergence holds promise creating smarter, safer, efficient, more secure systems. However, with surprising development quantum technology, most existing systems vulnerable to attacks. In this paper, we accordingly propose a lattice-based infrastructure for resistant, aiming...

Abstract New services, such as distributed photovoltaic regulation and control, pose new service requirements for networks in the power system. These include low latency, high reliability, large bandwidth. Consequently, heterogeneous communication face challenge of maintaining QoS while enhancing network resource utilization. Therefore, this paper puts forward a highly efficient optimization algorithm slicing scheduling networks. Our first step involves establishing an architectural...

Currently, the extension of end 5G power communication network by deploying multi-link multimedia access terminals integrating multiple technologies has become an important development trend in technology. This paper incorporates characteristics networks and designs a converged architecture based on 5G-MEC, aiming to support collaboration flexible edge resources. Based consideration constraints such as latency, computational resources, storage bandwidth this establishes joint optimization...

We develop a theoretical framework to study the training of federated learning models with partial gradients via over-the-air computing. The system consists an edge server and multiple clients, aiming collaboratively minimize global loss function. clients conduct local upload intermediate parameters (e.g. gradients) by analog transmissions. Specifically, each client modulates entries its gradient onto set common orthogonal waveforms sends out signal simultaneously server; owing limited...

A smart contract honeypot is a special type of contract. This seems to have obvious vulnerabilities in design. If user transfers certain amount funds the contract, then can withdraw However, once users try take advantage this seemingly vulnerability, they will fall into real trap. Consequently, user's investment cannot be retrieved. The induces other accounts launch funds, which seriously threatens security property on blockchain. Detection methods for honeypots are available. studying...

This paper aims to disclose the features of virus propagation in wireless sensor networks (WSNs). To this end, author analysed data transmission mode WSNs which follow IPv6 protocol. Then, such was simulated and discussed under different scenarios. Based on simulation results traditional model, a new model put forward describe WSNs. The proposed lays theoretical basis for relevant studies guarantee WSN security.

At present, image processing technology is more commonly applied to such fields as virtual reality and biological medicine, permeates rapidly through agricultural scientific research. The measuring method makes dairy cow type linear appraisal , which not only reduces workload but also gets rid of the shortcoming manual standard changing with human time disadvantage lacking impersonal impartiality. This paper describes a system for analyzing types based on processing, consists hardware...

Accurate hand joints detection from images is a fundamental topic which essential for many applications in computer vision and human interaction. This paper presents two stage network single unmarked image by using serial-parallel multi-scale feature fusion. In I, the regions are located pre-trained network, features of each detected region extracted shallow spatial representation module. The then fed into II, consists serially connected extraction modules with similar structures, called...

In fisheye images, rich distinct distortion patterns are regularly distributed in the image plane. These independent of visual content and provide informative cues for rectification. To make best such rectification cues, we introduce SimFIR, a simple framework based on self-supervised representation learning. Technically, first split into multiple patches extract their representations with Vision Transformer (ViT). learn fine-grained representations, then associate different specific model,...