Pinned Repositories
AdvCLIP
The implementation of our ACM MM 2023 paper "AdvCLIP: Downstream-agnostic Adversarial Examples in Multimodal Contrastive Learning"
AdvEncoder
The implementation of our ICCV 2023 paper "Downstream-agnostic Adversarial Examples"
AMT-GAN
[CVPR 2022] The official implementation of our CVPR 2022 paper "Protecting Facial Privacy: Generating Adversarial Identity Masks via Style-robust Makeup Transfer".
Android-Container
Method on running Linux containers (Docker) on the android platform. Migrate container from X86-Based ubuntu to ARM-Based android.
gECC
gECC: A GPU-based high-throughput framework for Elliptic Curve Cryptography
naturalcc
NaturalCC: An Open-Source Toolkit for Code Intelligence
SCVDT
Source Code Vulnerability Detection Tools（SCVDT）provides a vulnerable code database, vulnerability detection service for Java and C/C++ programs, and other security service.
VulCNN
VulDeePecker
VulDeePecker: A Deep Learning-Based System for Vulnerability Detection
YiTu
YiTu is an easy-to-use runtime to fully exploit the hybrid parallelism of different hardwares (e.g., GPU) to efficiently support the execution of various kinds of graph algorithms (e.g., GNNs).
CGCL-codes's Repositories
CGCL-codes/PathGraph
PathGraph, a path-centric graph processing system for fast iterative computation on large graphs with billions of edges. Large scale graph analysis applications typically involve datasets of massive scale. Most of existing approaches address the iterative graph computation problem by programming and executing graph computation using either vertex centric or edge centric approaches. We develop a path-centric graph processing system PathGraph for fast iterative computation on large graphs with billions of edges.
CGCL-codes/DGraph
DGraph is a system for directed graph processing with taking advantage of the strongly connected component structure. On this system, most graph partitions are able to reach convergence in order and need to be loaded into the main memory for exactly once, getting much lower data access cost and faster convergence.
CGCL-codes/TripeBit
TripeBit is designed based on two important observations. First, it is important to design a storage structure that can directly and efficiently query the RDF graph. This motivates us to design a compact storage and index structure in TripleBit. Second, in order to truly scale the RDF query processor, we need efficient index structures and query evaluation algorithms to minimize the size of intermediate results generated when evaluating queries, especially complex join queries. This leads us to the design decision that we should not only reduce the size of indexes, but also minimize the number of indexes used in query evaluation.
CGCL-codes/NightWatch
NightWatch is an extension of memory management system that provides general, transparent and low-overhead cache pollution control. NightWatch extends the memory mapping into two types: restrictive-mapping and open-mapping. The restrictive-mapping is used for restricting the pollution effect of the poor locality data, while the open-mapping is used for cache friendly data. When a malloc request arrives, NightWatch will predict the access locality of the to be allocated memory, determine the proper cache demand, and select the right mapping type for the malloc request. NightWatch is based on the observation that data within the same memory chunk or chunks within the same allocation context often share similar locality property. NightWatch embodies this observation by online monitoring current cache locality to predict future behavior and restricting potential cache polluters proactively.
CGCL-codes/ACStor
In virtualized data centers, the access of virtual disk images (VDIs) is critical for the overall system performance. As the system scales up to a large number of running VMs, the overall network traffic would become unbalanced with hot spots on some VMs inevitably, leading to I/O performance degradation when accessing the VMs. We propose an adaptive and collaborative VDI storage system (ACStor) to resolve the above performance issue, which can dynamically balance the traffic workloads in accessing VDI chunks based on the run-time network state.
CGCL-codes/LiveRender
LiveRender is an open source cloud gaming system based on graphics streaming. LiveRender intercepts the D3D graphics commands and migrates them from the server to the client. We use several compression techniques to reduce the data transmission of graphics streaming, and so LiveRender provides a better experience of cloud gaming.
CGCL-codes/Mammoth
Mammoth is a new MapReduce system which aims to improve MapReduce performance using global memory management. We have conducted extensive experiments with comparison against the native Hadoop platform. The results show that the Mammoth system can reduce the total job execution time by 40% in typical cases, without requiring any modifications of Hadoop programs. When a system is short of memory, the performance improvement can be up to 5 times as observed for CPU and I/O intensive applications, such as PageRank. Given the growing importance of supporting large-scale data processing and analysis, and the proven success of the MapReduce platform, the Mammoth system can have a promising potential and impact.
CGCL-codes/MURS
MURS is a memory scheduler for in-memory computing, which tries to mitigate the memory pressure for multiple data processing tasks sharing the executor.
CGCL-codes/FoG
FoG is a Fast Out-of-core Graph processing framework.
CGCL-codes/ParaCoM
Parallel core maintenance (ParaCoM) is an efficient parallel algorithm that can update the core numbers of vertices in a dynamic graph when multiple edges are inserted or deleted.
CGCL-codes/PBS
PBS is an efficient scheduling strategy in Xen VMM to accelerate various types of parallel application such as NPB-OpenMP, PARSEC, NPB-MPI.