A5101925381- Advanced Data Storage Technologies
- Caching and Content Delivery
- Distributed and Parallel Computing Systems
- Cloud Computing and Resource Management
- Peer-to-Peer Network Technologies
- Parallel Computing and Optimization Techniques
- Distributed systems and fault tolerance
- Advanced Graph Neural Networks
- Scientific Computing and Data Management
- Algorithms and Data Compression
- Topic Modeling
- Explainable Artificial Intelligence (XAI)
- Cloud Data Security Solutions
- Advanced Memory and Neural Computing
- Ferroelectric and Negative Capacitance Devices
- Domain Adaptation and Few-Shot Learning
- Technology and Security Systems
- Sentiment Analysis and Opinion Mining
- Data Stream Mining Techniques
- Wireless Sensor Networks and IoT
- Innovations in Concrete and Construction Materials
- Machine Learning and Data Classification
- Digital and Cyber Forensics
- Cellular Automata and Applications
- Big Data Technologies and Applications
Robert Bosch (Germany)
2023-2024
Wuhan National Laboratory for Optoelectronics
2013-2023
Huazhong University of Science and Technology
2013-2023
Sichuan International Studies University
2021
Tongji University
2021
Building Bridges
2011
Text classification is a fundamental and important area of natural language processing for assigning text into at least one predefined tag or category according to its content. Most the advanced systems are either too simple get high accuracy centered on using complex structures capture genuinely required information, which requires long time converge during their training stage. In order address such challenging issues, we propose dynamic embedding projection-gated convolutional neural...
Matrix computations are indispensable tools in science and engineering, while the electronic matrix suffered from limited speed. Alternatively, optical methods offered a high-speed solution. The computation is impractical unless it capable of extending to large scale, reconfiguring for general purpose operating with high efficiency. Here, we report experimentally demonstrate an computing processor based on integrated linear network. proposed photonic performing fundamental including XB = C,...
Efficient provenance storage is an essential step towards the adoption of provenance. In this paper, we analyze collected from multiple workloads with a view efficient storage. Based on our analysis, characterize properties respect to long term We then propose hybrid scheme that takes advantage graph structure data and inherent duplication in data. Our evaluation indicates scheme, combination web compression (adapted for provenance) dictionary encoding, provides best tradeoff terms ratio,...
Emerging Resistive Random Access Memory (ReRAM) is a promising candidate as the replacement for DRAM due to its low standby power, high density, scalability, and nonvolatility. By employing unique crossbar structure, ReRAM can be constructed with extremely density. However, faces some serious challenges in terms of performance, reliability, energy consumption. First, ReRAM’s structure causes an IR drop problem wire resistance sneak currents, which results nonuniform access latency banks...
Abstract Knowledge Graph Embedding (KGE) has attracted increasing attention. Relation patterns, such as symmetry and inversion, have received considerable focus. Among them, composition patterns are particularly important, they involve nearly all relations in KGs. However, prior KGE approaches often consider to be compositional only if well-represented the training data. Consequently, it can lead performance degradation, especially for under-represented patterns. To this end, we propose...
This paper analyzes the problems of load balance caused by data access in parallel file systems and gives an accurate way to estimate Data Servers. On base this evaluation, proposes adaptive loading migration (ALDM) strategy servers. Experiment results show ALDM has little impact on system performance during migration, improves about 10% after completed. Furthermore, also good adaptability various loads different storage systems.
Parity-based RAID provides system-level fault tolerance. However, parity updates caused by small writes introduce lots of extra I/Os, degrading I/O performance and wearing SSDs out. It has been proposed to use Non-Volatile Memory (NVM) as a cache on an SSD postpone until the whole stripe updated. this often fails because skewed distribution hot data chunks within stripe. In real workloads, it is difficult achieve full-stripe update even after long delay. paper, we propose Correlation aware...
Resistive Memory (ReRAM) is promising to be used as the storage-class memory due its good scalability, high density, low standby power and non-volatility. By employing unique crossbar structure, ReRAM can constructed with extremely density. However, ReRAM's structure suffers from an IR drop issue, which causes asymmetric access latency in arrays. Moreover, without transistors read write disturbances lead serious data reliability problem. In this paper, we propose Asymmetric-ReRAM (A-ReRAM)...
Parity based RAID suffers from poor small write performance due to heavy parity update overhead. The recently proposed method EPLOG constructs a new stripe with updated data chunks without updating old chunks. However, skewness of accesses, versions often need be kept protect other the same stripe. This seriously hurts efficiency recovering system device failures reconstructing preserved on failed devices.
Previous storage reliability solutions mainly include log files, snapshot, backup or ECC-based schemes. These technologies focus more on the whole disk system reliability, but not single data object. However, it is efficient to rebuild a file by recovering using above. In this paper, we propose use provenance, origin history of objects, damaged lost files. Provenance-based can exactly reconstruct right backtracking its generation process in past. The evaluation results show that...
Emerging Non-volatile Memory (NVM) is attractive because of its byte-addressability, durability, and DRAM-scale latency. Hashing indexes have been extensively used to provide fast query services in the storage system. Recent research proposes crash-consistent write-optimized hashing for NVM. However, existing NVM-based suffer from limited scalability when running on a Commercial Non-Volatile product, named Intel Optane DC Persistent Module (DCPMM), due bandwidth DCPMM. To achieve high load...
Efficient metadata management is a key issue in distributed file systems. Currently single server (MDS) was used to provide service many As result the MDS likely be bottleneck of system with expanding scale. To deal this problem, cluster has been advanced. There are subtree partitioning and hashing realize so on. In view strengths weaknesses previous methods, paper proposes half-dynamic strategy that supports replication metadata, dynamic addition failure recovery MDS. Finally we implement...
Typical load balancing strategies have been proved to be effective in traditional storage system that they succeeded providing specific algorithms on data partition and workload distribution. When comes OBS, an object based system, most of these are hard comply with the new feature- a smarter device named (OSD) is capable expanding upper burden release aware jobs from distributed file systems. While more intelligent powerful functions begin deployed OSD, such as replication technology, which...