Since the data samples on client devices are usually non-independent and non-identically distributed(non-IID),this will challenge the convergence of federated learning(FL)and reduce communication *** paper proposes Fe...
详细信息
Since the data samples on client devices are usually non-independent and non-identically distributed(non-IID),this will challenge the convergence of federated learning(FL)and reduce communication *** paper proposes FedQMIX,a node selection algorithm based on multi-agent reinforcement learning(MARL),to address these ***,we observe a connection between model weights and data distribution,and a clustering algorithm can group clients with similar data distribution into the same ***,we propose a QMIX-based mechanism that learns to select devices from clustering results in each communication round to maximize the reward,penalizing the use of more communication rounds and thereby improving the communication efficiency of ***,experiments show that FedQMIX can reduce the number of communication rounds by 11%and 30%on the MNIST and CIFAR-10 datasets,respectively,compared to the baseline algorithm(Favor).
The quality of photos is highly susceptible to severe weather such as heavy rain;it can also degrade the performance of various visual tasks like object *** removal is a challenging problem because rain streaks have d...
详细信息
The quality of photos is highly susceptible to severe weather such as heavy rain;it can also degrade the performance of various visual tasks like object *** removal is a challenging problem because rain streaks have different appearances even in one *** where rain accumulates appear foggy or misty,while rain streaks can be clearly seen in areas where rain is less *** propose removing various rain effects in pictures using a hybrid multiscale loss guided multiple feature fusion de-raining network(MSGMFFNet).Specially,to deal with rain streaks,our method generates a rain streak attention map,while preprocessing uses gamma correction and contrast enhancement to enhanced images to address the problem of rain *** these tools,the model can restore a result with abundant ***,a hybrid multiscale loss combining L1 loss and edge loss is used to guide the training process to pay attention to edge and content *** experiments conducted on both synthetic and real-world datasets demonstrate the effectiveness of our method.
Anomaly detection is an important task that improves the maturity and stability of a software during its development. System logs record rich information about the running states of the software and reveal key insight...
详细信息
This paper proposes a simple and efficient collision detection algorithm for a large number of moving objects. The basic idea is to minimise the number of moving objects that go through the complicated the collision c...
详细信息
This paper proposes a simple and efficient collision detection algorithm for a large number of moving objects. The basic idea is to minimise the number of moving objects that go through the complicated the collision checking process, which can improve the overall performance. To this end, we propose a visibility-based culling technique that identifies substantially small or hidden objects that do not cause any visual artifact even if we ignore them. This paper also develops a variable-size Morton codes to speed up the construction time of the Linear Bounding Volume Hierarchy (LBVH), which is used to efficiently check the proximity between objects efficiently. The visibility-based culling technique is based on the so called visibility map on the top of the g-Buffer technique. This map is a texture that contains the ID of the moving object in the screen space. The number of fragments of each object on the map is then counted in parallel manner on the GPU. If the number of fragments is less than a predefined threshold value, the algorithm does not include the object in the LBVH constructions step. Although the performance depends on the camera view point, we have verified through several experiments that the proposed algorithm improves the overall performance at least 70% even when the number of moving objects is more than 10,000.
With the emergence of AI for good, there has been an increasing interest in building computer vision data-driven deep learning inclusive AI solutions. Sign language Recognition (SLR) has gained attention recently. It ...
详细信息
The cloud network is rapidly growing due to a massive increase in interconnected devices and the emergence of different technologies such as the Internet of things, fog computing, and artificial intelligence. In respo...
详细信息
We propose DyGFormer, a new Transformer-based architecture for dynamic graph learning. DyGFormer is conceptually simple and only needs to learn from nodes' historical first-hop interactions by: (i) a neighbor co-o...
We propose DyGFormer, a new Transformer-based architecture for dynamic graph learning. DyGFormer is conceptually simple and only needs to learn from nodes' historical first-hop interactions by: (i) a neighbor co-occurrence encoding scheme that explores the correlations of the source node and destination node based on their historical sequences; (ii) a patching technique that divides each sequence into multiple patches and feeds them to Transformer, allowing the model to effectively and efficiently benefit from longer histories. We also introduce DyGLib, a unified library with standard training pipelines, extensible coding interfaces, and comprehensive evaluating protocols to promote reproducible, scalable, and credible dynamic graph learning research. By performing exhaustive experiments on thirteen datasets for dynamic link prediction and dynamic node classification tasks, we find that DyGFormer achieves state-of-the-art performance on most of the datasets, demonstrating its effectiveness in capturing nodes' correlations and long-term temporal dependencies. Moreover, some results of baselines are inconsistent with previous reports, which may be caused by their diverse but less rigorous implementations, showing the importance of DyGLib. All the used resources are publicly available at https://***/yule-BUAA/DyGLib.
Compiler error recovery diagnostics facilitates software development as it provides the possible causes and suggestions on potential programming errors. However, due to compiler bugs, error recovery diagnostics could ...
详细信息
Compiler error recovery diagnostics facilitates software development as it provides the possible causes and suggestions on potential programming errors. However, due to compiler bugs, error recovery diagnostics could be erroneous, spurious, missing, or even crashing for mature production compilers like GCC and Clang. Compiler testing is one of the most widely used ways of ensuring its quality. However, existing compiler diagnostics testing approaches (e.g., DIPROM) only consider the typically syntactically valid test programs as inputs, which are unlikely to trigger compiler error recovery defects. Therefore, in this paper, we propose the first mutation based approach for Compiler Error Recovery diagnostics Testing, called CERTest. Specifically, CERTest first explores the mutation space for a given seed program, and leverages a series of mutation configurations (which are referred as a series of mutators applying for a seed) to iteratively mutate the structures of the seed, so as to generate error-sensitive program variants for triggering compiler error recovery mechanisms. To effectively construct error-sensitive structures, CERTest then applies a novel furthest-first based selection approach to select a set of representative mutation configurations to generate program variants in each iteration. With the generated program variants, CERTest finally leverages differential testing to detect error recovery defects in different compilers. The experiments on GCC and Clang demonstrate that CERTest outperforms five state-of-the-art approaches (i.e., DIPROM, Ccoft, Clang-fuzzer, AFL++, and HiCOND) by up to 13.10%similar to 221.61% on average in the term of bug-finding capability, and CERTest detects 9 new error recovery defects, 5 of which have been confirmed or fixed by developers.
The growing prevalence of text reuse and plagiarism in various fields has led to an urgent need for reliable computational methods for detection. However, current commercial plagiarism detection systems are ineffectiv...
详细信息
The field of architecture,engineering,and construction(AEC)is continually striving to use resources efficiently and manage complex *** more than ever,there is a strong need for digital transformation in *** improvemen...
详细信息
The field of architecture,engineering,and construction(AEC)is continually striving to use resources efficiently and manage complex *** more than ever,there is a strong need for digital transformation in *** improvement in the accessibility of consumer-based head-mounted displays(HMD)is attracting different entertainment and research fields to immersive virtual reality(VR)*** Information Modeling(BIM)is known as a promising technology in *** full potential of BIM is not yet employed to empower this field,however,and this could be a result of some barriers still to be surmounted by BIM in both technological and management *** of these barriers is the communication and collaboration between design,construction,operation,and maintenance *** can fill this gap by providing additional capabilities for BIM which either were not available before or were not possible to employ in practical *** this paper,we systematically review recent research around the application of VR in BIM and discuss the results using the PRISMA *** discuss the most commonly used technologies,software,and evaluation methods and the various applications of VR in the reviewed ***,we extend the discussion by summarizing the potential future work in this area.
暂无评论