A5049298612- Distributed and Parallel Computing Systems
- Service-Oriented Architecture and Web Services
- Business Process Modeling and Analysis
- Cloud Computing and Resource Management
- Caching and Content Delivery
- Advanced Data Storage Technologies
- Distributed systems and fault tolerance
- Peer-to-Peer Network Technologies
- Scientific Computing and Data Management
- IoT and Edge/Fog Computing
- Graph Theory and Algorithms
- Advanced Graph Neural Networks
- Gastrointestinal Bleeding Diagnosis and Treatment
- Parallel Computing and Optimization Techniques
- Image and Video Quality Assessment
- Opportunistic and Delay-Tolerant Networks
- Stochastic Gradient Optimization Techniques
- Context-Aware Activity Recognition Systems
- Advanced Neural Network Applications
- Semantic Web and Ontologies
- Network Traffic and Congestion Control
- Complexity and Algorithms in Graphs
- Gastrointestinal disorders and treatments
- Nanoparticle-Based Drug Delivery
- Brain Tumor Detection and Classification
Shanghai Jiao Tong University
2021-2024
Tsinghua University
2015-2024
National Engineering Research Center for Nanotechnology
2024
Center for Information Technology
2018
National Supercomputing Center in Wuxi
2018
Beihang University
2018
In recent years, the deep integration of basic research and clinical translational nanotechnology oncology has led to emergence a new branch, namely integrated nano-oncology. This is an emerging important interdisciplinary field, which plays irreplaceable role in diagnosis, treatment, early warning, monitoring prevention tumors, become frontier. Here main advances nano-oncology was reviewed, mainly included controlled preparation nanomaterials, ultra-sensitive detection tumor biomarkers,...
A virtual machine (VM) has been serving as a crucial component in cloud computing with its rich set of convenient features. The high overhead VM well addressed by hardware support such Intel virtualization technology (VT), and improvement recent hypervisor implementation Xen, KVM, etc. However, the demand on image storage remains challenging problem. Existing systems have made efforts to reduce consumption means deduplication within area network (SAN) cluster. Nevertheless, an SAN cannot...
The efficiency of distributed shared memory (DSM) has been greatly improved by recent hardware technologies. But, the difficulty management can still be a major obstacle to democratization DSM, especially when partial failure participating clients (e.g., due crashed processes or machines) should tolerated.
This paper reports our efforts on swCaffe, a highly efficient parallel framework for accelerating deep neural networks (DNNs) training Sunway TaihuLight, the current fastest supercomputer in world that adopts unique many-core heterogeneous architecture, with 40,960 SW26010 processors connected through customized communication network.First, we point out some insightful principles to fully exploit performance of innovative architecture.Second, propose set optimization strategies redesigning...
As a widely used programming model and implementation for processing large data sets, MapReduce performs poorly on heterogeneous clusters, which, unfortunately, are common in current computing environments. To deal with the problem, this paper: 1) analyzes causes of performance degradation identifies key one as volume inter-node transfer resulted from even distribution among nodes different capabilities, 2) proposes ActCap, solution that uses Markov chain based to do node-capability-aware...
Many real-world graphs are temporal in nature, where the information indicates when a particular edge is changed (e.g., insertion and deletion). Performing random walks on such of paramount value. The state-of-the-art sampling strategies tailored for conventional static thus cannot effectively tackle dynamic nature due to several significant efficiency challenges, i.e., high complexity, gigantic index space, poor programmability.
Recently, the Storage Networking Industry Association (SNIA) has released first standard for cloud interoperability. With more and standards interoperability emerging, it can be expected that a global resource exchange market will form. In such market, is challenging to present dynamic pricing scheme meet different requirements. To cope with challenge, in this paper we framework constructing markets then propose knowledge-based continuous double auction (CDA) model determines price of...
Serverless computing is renowned for its computation elasticity, yet full potential often constrained by the requirement functions to operate within local and dedicated background environments, resulting in limited memory elasticity. To address this limitation, paper introduces TrEnv, a co-designed integration of serverless platform with operating system CXL/RDMA-based remote pools two key areas. Firstly, TrEnv repurposable sandboxes, which can be shared across different hence, substantially...
Dynamic programming (DP) is an effective technique for many search and optimization problems. However, the high arithmetic complexity limits its extensive use. Although modern processor architectures with multiple cores SIMD (single instruction data) instructions provide increasingly computing power, even state-of-the-art fully optimized algorithm still largely underutilizes multi-core processors. In this paper we propose to improve one family of DP, nonserial polyadic DP (NPDP), targeting a...
Out-of-core random walk system has recently attracted a lot of attention as an economical way to run billions walkers over large graphs. However, existing out-of-core systems are all built upon general graph processing frameworks, and hence do not take advantage the unique properties applications. Different from traditional analysis algorithms, sampling process can be decoupled walkers. It enables reserve only pre-sample results in memory, which typically much smaller than entire edge set....
Capsule endoscopy (CE) enables noninvasive examination of the gastrointestinal (GI) tract. However, there is still no viable way to solve capsule retention (CR), which most worrisome complication CE in a clinic. The CR rate did not decrease with continuous development CE. Here, we proposed novel approach that fabricated programmable adaptive anti-retention module (PARM) small size and integrated it into capsules alleviate GI We developed durable, ultrawide range highly sensitive pressure...
MapReduce, as a programming model and implementation for processing large data sets on clusters with hundreds or thousands of nodes, has gained wide adoption. In spite the fact, we found that MapReduce commodity clusters, which are usually equipped limited memory hard-disk drive (HDD) have processors multiple many cores, does not scale expected number processor cores increases. The key reason this is underlying low-speed HDD storage cannot meet requirement frequent IO operations. Though...
Context awareness, context sharing and processing are key requirements for the future CSCW, HCI ubiquitous computing systems. Until now, collaborative factors have been seldom specifically addressed. This paper argues that a generic model is very important building context-aware applications. A new semantic rich environment proposed. The conceptual described as ontology contextual applications (OCCA). query & memory service, matching service control policies illustrated. Information space,...
As the device holding great promise to realize vision of pervasive computing, smartphones with Internet access and video playing capability have become commercial off shelf products today. one most popular applications on smartphones, online playing, however, still faces some problems. From perspective service providers, they provide files different formats qualities meet diverse needs various consumers. consumers, if or more codecs required open a file supplied by certain provider are not...