检索结果-内蒙古大学图书馆

ScalabFS: A scalable BFS accelerator on HBM-enhanced FPGAs

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Li, Kexin Liu, Chenhao Shao, Zhiyuan Wang, Zeke Wu, Minkang Chen, Jiajie Liao, Xiaofei Jin, Hai National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab Cluster and Grid Computing Lab School of Computer Science and Technology Huazhong University of Science and Technology Wuhan30074 China Collaborative Innovation Center of Artificial Intelligence Zhejiang University China

High Bandwidth Memory (HBM) provides massive aggregated memory bandwidth by exposing multiple memory channels to the processing units. To achieve high performance, an accelerator built on top of an FPGA configured with HBM (i.e., FPGA-HBM platform) needs to scale its performance according to the available memory channels. In this paper, we propose an accelerator for BFS (Breadth-First Search) algorithm, named as ScalabFS, that builds multiple processing elements to sufficiently exploit the high bandwidth of HBM to improve efficiency. We implement the prototype system of ScalabFS and conduct BFS in both real-world and synthetic scale-free graphs on Xilinx Alveo U280 FPGA card real hardware. The experimental results show that ScalabFS scales its performance almost linearly according to the available memory pseudo channels (PCs) from the HBM2 subsystem of U280. By fully using the 32 PCs and building 64 processing elements (PEs) on U280, ScalabFS achieves a performance up to 19.7 GTEPS (Giga Traversed Edges Per Second). When conducting BFS in sparse real-world graphs, ScalabFS achieves equivalent GTEPS to Gunrock running on the state-of-art Nvidia V100 GPU that features 64-PC HBM2 (twice memory bandwidth than U280). Copyright © 2021, The Authors. All rights reserved.

关键词： Field programmable gate arrays (FPGA)

CKG: Dynamic Representation Based on Context and Knowledge Graph

学校读者我要写书评

暂无评论

CKG: Dynamic Representation Based on Context and Knowledge G...

International Conference on Pattern Recognition

作者： Xunzhu Tang Tiezhu Sun Rujie Zhu Shi Wang National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab Cluster and Grid Computing Lab School of Computer Science and Technology Huazhong University of Science and Technology Wuhan China Momenta Suzhou China University of Central Florida Orlando FL USA Institute of Computing Technology Chinese Academy beijing China

Recently, neural language representation models pre-trained on large corpus can capture rich co-occurrence information and be fine-tuned in downstream tasks to improve the performance. As a result, they have achieved state-of-the-art results in a large range of language tasks. However, there exists other valuable semantic information such as similar, opposite, or other possible meanings in external knowledge graphs (KGs). We argue that entities in KGs could be used to enhance the correct semantic meaning of language sentences. In this paper, we propose a new method CKG: Dynamic Representation Based on Context and Knowledge Graph. On the one side, CKG can extract rich semantic information of large corpus. On the other side, it can make full use of inside information such as co-occurrence in large corpus and outside information such as similar entities in KGs. We conduct extensive experiments on a wide range of tasks, including QQP, MRPC, SST-5, SQuAD, CoNLL 2003, and SNLI. The experiment results show that CKG achieves SOTA 89.2 on SQuAD compared with SAN (84.4), ELMo (85.8), and BERTBase (88.5).

关键词： Semantics data mining Task analysis Context modeling

Bill-EVR: An Embodied Virtual Reality Framework for Reward-and-Error-Based Motor Rehab-Learning

学校读者我要写书评

暂无评论

Bill-EVR: An Embodied Virtual Reality Framework for Reward-a...

IEEE International Conference on Rehabilitation Robotics (ICORR)

作者： Federico Nardi Shlomi Haar A.Aldo Faisal UKRI Centre for Doctoral Training in AI for Healthcare Imperial College London London United Kingdom Dept. of Computing & Dept. of Bioengineering Brain & Behaviour Lab Imperial College London London United Kingdom Care Research & Technology centre UK Dementia Research Institute Department of Brain Sciences Imperial College London London United Kingdom Chair in Digital Health & Data Science University of Bayreuth Bayreuth Germany

VR rehabilitation is an established field by now, however, it often refers to computer screen-based interactive rehabilitation activities. In recent years, there was an increased use of VR-headsets, which can provide an immersive virtual environment for real-world tasks, but they are lacking any physical interaction with the task objects and any proprioceptive feedback. Here, we focus on Embodied Virtual Reality (EVR), an emerging field where not only the visual input via VR-headset but also the haptic feedback is physically correct. This happens because subjects interact with physical objects that are veridically aligned in Virtual Reality. This technology lets us manipulate motor performance and motor learning through visual feedback perturbations. Bill-EVR is a framework that allows interventions in the performance of real-world tasks, such as playing pool billiard, engaging end-users in motivating life-like situations to trigger motor (re)learning - subjects see in VR and handle the real-world cue stick, the pool table and shoot physical balls. Specifically, we developed our platform to isolate and evaluate different mechanisms of motor learning to investigate its two main components, error-based and reward-based motor adaptation. This understanding can provide insights for improvements in neurorehabilitation: indeed, reward-based mechanisms are putatively impaired by degradation of the dopaminergic system, such as in Parkinson's disease, while error-based mechanisms are essential for recovering from stroke-induced movement errors. Due to its fully customisable features, our EVR framework can be used to facilitate the improvement of several conditions, providing a valid extension of VR-based implementations and constituting a motor learning tool that can be completely tailored to the individual needs of patients.

关键词：

Moto: Enhancing Embedding with Multiple Joint Factors for Chinese Text Classification

学校读者我要写书评

暂无评论

Moto: Enhancing Embedding with Multiple Joint Factors for Ch...

International Conference on Pattern Recognition

作者： Xunzhu Tang Rujie Zhu Tiezhu Sun Shi Wang National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab Cluster and Grid Computing Lab School of Computer Science and Technology Huazhong University of Science and Technology Wuhan China University of Central Florida Orlando FL USA Momenta Suzhou China Institute of Computing Technology Chinese Academy beijing China

Recently, language representation techniques have achieved great performances in text classification. However, most existing representation models are specifically designed for English materials, which may fail in Chinese because of the huge difference between these two languages. Actually, few existing methods for Chinese text classification process texts at a single level. However, as a special kind of hieroglyphics, radicals of Chinese characters are good semantic carriers. In addition, Pinyin codes carry the semantic of tones, and Wubi reflects the stroke structure information, etc. Unfortunately, previous researches neglected to find an effective way to distill the useful parts of these four factors and to fuse them. In our works, we propose a novel model called Moto: Enhancing Embedding with Multiple Joint Factors. Specifically, we design an attention mechanism to distill the useful parts by fusing the four-level information above more effectively. We conduct extensive experiments on four popular tasks. The empirical results show that our Moto achieves SOTA 0.8316 (F₁-score, 2.11% improvement) on Chinese news titles, 96.38 (1.24% improvement) on Fudan Corpus and 0.9633 (3.26% improvement) on THUCNews.

关键词： Fuses Text categorization Semantics Pattern recognition Task analysis

One-Shot Medical Landmark Localization by Edge-Guided Transform and Noisy Landmark Refinement

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Yin, Zihao Gong, Ping Wang, Chunyu Yu, Yizhou Wang, Yizhou Center for Data Science Peking University Beijing China Deepwise AI Lab Beijing China Microsoft Research Asia Beijing China The University of Hong Kong Hong Kong Center on Frontiers of Computing Studies School of Computer Science Peking University Beijing China Inst. for Artificial Intelligence Peking University Beijing China

As an important upstream task for many medical applications, supervised landmark localization still requires non-negligible annotation costs to achieve desirable performance. Besides, due to cumbersome collection procedures, the limited size of medical landmark datasets impacts the effectiveness of large-scale self-supervised pre-training methods. To address these challenges, we propose a two-stage framework for one-shot medical landmark localization, which first infers landmarks by unsupervised registration from the labeled exemplar to unlabeled targets, and then utilizes these noisy pseudo labels to train robust detectors. To handle the significant structure variations, we learn an end-to-end cascade of global alignment and local deformations, under the guidance of novel loss functions which incorporate edge information. In stage II, we explore self-consistency for selecting reliable pseudo labels and cross-consistency for semi-supervised learning. Our method achieves state-of-the-art performances on public datasets of different body parts, which demonstrates its general applicability. Copyright © 2022, The Authors. All rights reserved.

关键词： Medical applications

iGniter: Interference-Aware GPU Resource Provisioning for Predictable DNN Inference in the Cloud

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Xu, Fei Xu, Jianian Chen, Jiabin Chen, Li Shang, Ruitao Zhou, Zhi Liu, Fangming The Shanghai Key Laboratory of Multidimensional Information Processing School of Computer Science and Technology East China Normal University 3663 N. Zhongshan Road Shanghai200062 China The School of Computing and Informatics University of Louisiana at Lafayette 301 East Lewis Street LafayetteLA70504 United States The Guangdong Key Laboratory of Big Data Analysis and Processing School of Computer Science and Engineering Sun Yat-sen University 132 E. Waihuan Road Guangzhou510006 China The National Engineering Research Center for Big Data Technology and System The Services Computing Technology and System Lab Cluster and Grid Computing Lab School of Computer Science and Technology Huazhong University of Science and Technology 1037 Luoyu Road Wuhan430074 China

GPUs are essential to accelerating the latency-sensitive deep neural network (DNN) inference workloads in cloud datacenters. To fully utilize GPU resources, spatial sharing of GPUs among co-located DNN inference workloads becomes increasingly compelling. However, GPU sharing inevitably brings severe performance interference among co-located inference workloads, as motivated by an empirical measurement study of DNN inference on EC2 GPU instances. While existing works on guaranteeing inference performance service level objectives (SLOs) focus on either temporal sharing of GPUs or reactive GPU resource scaling and inference migration techniques, how to proactively mitigate such severe performance interference has received comparatively little attention. In this paper, we propose iGniter, an interference-aware GPU resource provisioning framework for cost-efficiently achieving predictable DNN inference in the cloud. iGniter is comprised of two key components: (1) a lightweight DNN inference performance model, which leverages the system and workload metrics that are practically accessible to capture the performance interference;(2) A cost-efficient GPU resource provisioning strategy that jointly optimizes the GPU resource allocation and adaptive batching based on our inference performance model, with the aim of achieving predictable performance of DNN inference workloads. We implement a prototype of iGniter based on the NVIDIA Triton inference server hosted on EC2 GPU instances. Extensive prototype experiments on four representative DNN models and datasets demonstrate that iGniter can guarantee the performance SLOs of DNN inference workloads with practically acceptable runtime overhead, while saving the monetary cost by up to 25% in comparison to the state-of-the-art GPU resource provisioning strategies. Copyright © 2022, The Authors. All rights reserved.

关键词： Deep neural networks

Toward the Understanding of Deep Text Matching Models for Information Retrieval

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Chen, Lijuan Lan, Yanyan Pang, Liang Guo, Jiafeng Cheng, Xueqi Sogou Inc. Beijing China Institute for AI Industry Research Tsinghua University Beijing China Data Intelligence System Research Center Institute of Computing Technology Chinese Academy of Sciences Beijing China CAS Key Lab of Network Data Science and Technology Institute of Computing Technology Chinese Academy of Sciences Beijing China

Semantic text matching is a critical problem in information retrieval. Recently, deep learning techniques have been widely used in this area and obtained significant performance improvements. However, most models are black boxes and it is hard to understand what happened in the matching process, due to the poor interpretability of deep learning. This paper aims at tackling this problem. The key idea is to test whether existing deep text matching methods satisfy some fundamental heuristics in information retrieval. Specifically, four heuristics are used in our study, i.e., term frequency constraint, term discrimination constraint, length normalization constraints, and TF-length constraint. Since deep matching models usually contain many parameters, it is difficult to conduct a theoretical study for these complicated functions. In this paper, We propose an empirical testing method. Specifically, We first construct some queries and documents to make them satisfy the assumption in a constraint, and then test to which extend a deep text matching model trained on the original dataset satisfy the corresponding constraint. Besides, a famous attribution based interpretation method, namely integrated gradient, is adopted to conduct detailed analysis and guide for feasible improvement. Experimental results on LETOR 4.0 and MS Marco show that all the investigated deep text matching methods, both representation and interaction based methods, satisfy the above constraints with high probabilities in statistics. We further extend these constraints to the semantic settings, which are shown to be better satisfied for all the deep text matching models. These empirical findings give clear understandings on why deep text matching models usually perform well in information retrieval. We believe the proposed evaluation methodology will be useful for testing future deep text matching models. © 2021, CC BY.

关键词： Deep learning

Explainable Unsupervised Multi-Sensor Industrial Anomaly Detection and Categorization

学校读者我要写书评

暂无评论

Explainable Unsupervised Multi-Sensor Industrial Anomaly Det...

International Conference on Machine Learning and Applications (ICMLA)

作者： Mina Ameli Philipp Aaron Becker Katharina Lankers Markus van Ackeren Holger Bähring Wolfgang Maaß Cognitive Assistants Systems Department DFKI GmbH Saarbrücken Germany Lab Department deZem GmbH Berlin Germany Research & Development – Data Science Department Schott AG Mainz Germany Development Department SEITEC GmbH Erfurt Germany Smart Service Engineering Department DFKI GmbH Saarbrücken Germany

Real-time Anomaly Detection is of great importance in industrial applications in order to have high-quality production and avoid downtime or failure of the system. In this paper, we study the application of anomaly detection over the multivariate data collected from Glass Production Industry. Our experiments utilize and compare different Unsupervised multivariate time series Anomaly Detection and Localization algorithms that have already demonstrated significant results on the state-of-the-art data sets. We propose a two-level multivariate anomaly detection approach that not only detects anomalous events in the production line but also categorize the different type of anomalies based on statistical pattern recognition. Furthermore, we localize the anomalous sensors by utilizing Explainable-AI approaches to help better decision-making in glass production monitoring. In this work, we propose an efficient pipeline for Anomaly Detection, Categorization and Localization which the experiments show promising results.

关键词： Location awareness Machine learning algorithms Time series analysis Pipelines Production Glass Real-time systems

LCCG: A Locality-Centric Hardware Accelerator for High Throughput of Concurrent Graph Processing

学校读者我要写书评

暂无评论

LCCG: A Locality-Centric Hardware Accelerator for High Throu...

Supercomputing Conference

作者： Jin Zhao Yu Zhang Xiaofei Liao Liaang He Binzsheng He Haikun Liu Hai Jin Huazhong University of Science and Technology China National Engineering Research Center for Big Data Technology and System Services Computing Technology and System Lab Cluster and Grid Computing Lab School of Computer Science and Technology Huazhong University of Science and Technology Wuhan China University of Warwick United Kingdom National University of Singapore Singapore

ISBN: (数字)9781450384421

ISBN: (纸本)9781665483902

In modern data centers, massive concurrent graph processing jobs are being processed on large graphs. However, existing hardware/-software solutions suffer from irregular graph traversal and intense resource contention. In this paper, we propose LCCG, a Locality-entric programmable accelerator that augments the many-core processor for achieving higher throughput of Concurrent Graph processing jobs. Specifically, we develop a novel topology-aware execution approach into the accelerator design to regularize the graph traversals for multiple jobs on-the-fly according to the graph topology, which is able to fully consolidate the graph data accesses from concurrent jobs. By reusing the same graph data among more jobs and coalescing the accesses of the vertices' states for these jobs, LCCG can improve the core utilization. We conduct extensive experiments on a simulated 64-core processor. The results show that LCCG improves the throughput of the cutting-edge software system by 11.3~23.9 times with only 0.5% additional area cost. More-over, LCCG gains the speedups of 4.7~10.3, 5.5~13.2, and 3.8~8.4 times over state-of-the-art hardware graph processing accelerators (namely, HATS, Minnow, and PHI, respectively).

关键词： data centers Costs High performance computing Throughput Software systems Topology Hardware acceleration