image classification is one of the most important fundamental research topics in machine learning and imageprocessing. Recently, hypergraph learning, which can model the high-order relationship of samples and fusion ...
详细信息
ISBN:
(纸本)9781538666142
image classification is one of the most important fundamental research topics in machine learning and imageprocessing. Recently, hypergraph learning, which can model the high-order relationship of samples and fusion multimodal features, has received the attention of many researchers. However, existing multimodal hypergraph learning methods face two problems, i.e., how to construct hyperedges and how to determine the weights of hyperedges. this paper proposes an adaptive multimodal hypergraph learning method (AMH) to address these two challenges. AMH uses multiple neighborhoods method to avoid generating a k-uniform hyperedge, and optimizes the weights withthe penalty function method to take the initial labels into consideration. the experimental results demonstrate the effectiveness of AMH compared withthe stateof-the-art methods.
Convolutional neural networks have existed for many years, but recently they have been developed to a greater depth and widththan ever before withthe increase in the computing power of graphics processing units. Con...
详细信息
ISBN:
(纸本)9781538685341
Convolutional neural networks have existed for many years, but recently they have been developed to a greater depth and widththan ever before withthe increase in the computing power of graphics processing units. Convolutional neural networks are widely used in a variety of artificial intelligence applications, including in manufacturing, agriculture, and medicine. the use of artificial intelligence in various industrial fields is expected to increase. However, improvements in network training efficiency have not resulted in a reciprocal improvement in computational power for identification applications. this paper proposes several types of neural networks that are based on well-known networks such as AlexNet, GoogleNet, and ResNet, whose characteristics have been captured and implemented in lower layer neural networks. From the experimental results, using these hybrid neural networks can bring improved accuracy, with well optimized computational time costs compared to networks that require a large amount of computation.
We present a new greedy b-MATCHING algorithm suitable for running on a GPU. Our algorithm differs from previous efforts at designing parallel algorithms for this problem in that it does not use software locks and that...
详细信息
ISBN:
(纸本)9781538655559
We present a new greedy b-MATCHING algorithm suitable for running on a GPU. Our algorithm differs from previous efforts at designing parallel algorithms for this problem in that it does not use software locks and that it also exploits substantially more of the available concurrency. We achieve this by allowing the same vertex to concurrently match with several other vertices and also by letting multiple vertices simultaneously match withthe same target vertex. We have compared our algorithm using a Pascal P100 GPU withthe previous best shared memory algorithm for this problem both when running on a 16 core Xeon E5 and on a Xeon Phi. On average our algorithm outperforms the Xeon E5 by a factor of 4.6 and the Xeon Phi by a factor of 2.3. We also show that our algorithm using an NVIDIA DGX-1 multi-GPU system is highly competitive compared to a distributed memory implementation running on one of the top ten computers from the current TOP500 list.
Due to the growth of data scale, distributed machine learning has become more important than ever. Some recent work, like TuX(2), show promising prospect in dealing with distributed machine learning by leveraging the ...
详细信息
ISBN:
(纸本)9781728111414
Due to the growth of data scale, distributed machine learning has become more important than ever. Some recent work, like TuX(2), show promising prospect in dealing with distributed machine learning by leveraging the power of graph computation, but still leave some key problems unsolved. In this paper, we propose Cymbalo, a new distributed graph processing framework for large-scale machine learning algorithms. To satisfy the specific characteristics of machine learning, Cymbalo employs a heterogeneity-aware data model, a hybrid computing model and a vector-aware programming model, to ensure small memory footprint, good computation efficiency and expressiveness. the experiment results show that Cymbalo outperforms Spark by 2.4x-3.2x, and PowerGraph by up to 5.8x. Moreover, Cymbalo can also outperform Angel, a recent parameter server system, by 1.6x-2.1x.
Real-time road congestion detection allows improving traffic safety and route planning. In this work, we propose to use streaming graph processingalgorithms for road congestion detection and evaluate their accuracy a...
详细信息
ISBN:
(纸本)9781728111414
Real-time road congestion detection allows improving traffic safety and route planning. In this work, we propose to use streaming graph processingalgorithms for road congestion detection and evaluate their accuracy and performance. We represent road infrastructure sensors in the form of a directed weighted graph and adapt the Connected Components algorithm and some existing graph processingalgorithms, originally used for community detection in social network graphs, for the task of road congestion detection. In our approach, we detect Connected Components or communities of sensors with similarly weighted edges that reflect different states in the traffic, e.g., free flow or congested state, in regions covered by detected sensor groups. We have adapted and implemented the Connected Components and community detection algorithms for detecting groups in the weighted sensor graphs in batch and streaming manner. We evaluate our approach by building and processingthe road infrastructure sensor graph for Stockholm's highways using real-world data from the Motorway Control System operated by the Swedish traffic authority. Our results indicate that the Connected Components and DenGraph community detection algorithms can detect congestion with accuracy up to approximate to 94% for Connected Components and up to approximate to 88% for DenGraph. the Louvain Modularity algorithm for community detection fails to detect congestion regions for sparsely connected graphs, representing roads that we have considered in this study. the Hierarchical Clustering algorithm using speed and density readings is able to detect congestion without details, such as shockwaves.
In this paper, we propose a blind/no-reference 3D image quality assessment scheme that utilizes binocular visual characteristics. the design of this scheme is motivated by studies on the perception of distorted stereo...
详细信息
In the paper an introductory research on visual system for underwater scene change detection and environment monitoring by an autonomous underwater drone is described. the systems sensor front-end is composed of a sid...
详细信息
ISBN:
(纸本)9781728111414
In the paper an introductory research on visual system for underwater scene change detection and environment monitoring by an autonomous underwater drone is described. the systems sensor front-end is composed of a side-scan sonar, a set of video cameras and lighting module. the system contains a number of processing blocks. First is responsible for signal filtering and conditioning. the main processing unit is based on a change detection module operating with our tensor based scene change detection unit. thanks to our developed parallel algorithm for tensor based model construction, the system is able to find abrupt scene changes, as well as presence of previously unseen objects which can be of interest and which are left for further monitored. In the work-in-progress report the system architecture, theoretical foundations, as well as preliminary experimental results are presented.
In mobile wireless multimedia sensor networks (MWMSNs) image compression task coordination, the existing methods without considering the dynamic changes of the processing ability and location of cooperative nodes, whi...
详细信息
ISBN:
(纸本)9781450359528
In mobile wireless multimedia sensor networks (MWMSNs) image compression task coordination, the existing methods without considering the dynamic changes of the processing ability and location of cooperative nodes, which will cause frequent interruptions of image compression tasks and task data re-transmission. To solve these problems, an image compression task cooperation algorithm based on dynamic alliance (ATDA) is proposed. the image compression task is divided into image transfer subtask and image compression subtask based on the task stable execution time and the principles and constraints of task decomposition. Simulation results show that the proposed algorithm can realize the task load balancing of the alliance collaboration nodes and reduce the execution time of the image compression task and energy consumption.
Red blood cell segmentation in microscopic images is the first step for various clinical studies carried out on blood samples such as cell counting, cell shape identification, etc. Conventional methods while often sho...
详细信息
ISBN:
(纸本)9781728111414
Red blood cell segmentation in microscopic images is the first step for various clinical studies carried out on blood samples such as cell counting, cell shape identification, etc. Conventional methods while often showing a high accuracy are heavily depending on the acquisition modality. Deep learning approaches have shown to be more robust regarding such modalities and still showing a comparable accuracy. In this paper, we first investigate necessary steps to apply a specific type of deep learning methods, namely fully convolutional networks, to red blood cell segmentation. Based on data given and constraints imposed by our partners mainly regarding a high throughput of their data we then describe an exemplary application. First results show, that even with a focus on high performance a good accuracy above 90% can be reached.
this paper presents a particle -swarm -optimized 2D Otsu segmentation algorithm based on fractional order differential and cloud model (hereinafter called the TOCPSO) to solve the low efficiency of the traditional 2D ...
详细信息
ISBN:
(纸本)9781728111414
this paper presents a particle -swarm -optimized 2D Otsu segmentation algorithm based on fractional order differential and cloud model (hereinafter called the TOCPSO) to solve the low efficiency of the traditional 2D Otsu segmentation algorithm. the proposed algorithm uses 2D Otsu as fitness function, and adaptively changes the fractional order according to the state of particle swarm. It updates the speed and position values by obtaining the a -order derivative at the current moment. Based on the fitness value, the particles are divided into different populations. Experiments show that the proposed segmentation process improves the convergence speed of the algorithm under the premise of guaranteeing the segmentation effect.
暂无评论