Consistency learning combined with pseudo label training is a mainstream paradigm in semi-supervised medical image segmentation (SSMIS), yet it faces two challenges: 1)The one-sided search and handling of uncertainty regions make it difficult to address complex uncertainty situations caused by multiple networks. 2)The premature reaching of consensus in traditional SSMIS frameworks hinders the information complementarity between sub-networks. To address these, we propose a multi-view interactive interference training framework based on cross-network uncertainty (MIICU). Specifically, we design a cross-network uncertainty region searching (CUS) module, which combines the predictions from two networks to provide more reasonable uncertainty regions. Then, we design a dual-decision complementary displacement (DCD) strategy that performs displacement operations on cross-network uncertainty regions in different scenarios, so as to facilitate the learning of these regions. We further propose a multi-view interactive interference training strategy by expanding the training environment with three different concatenation forms, encouraging information complementarity between sub-networks. Evaluation on the TN3K and ACDC datasets shows that our approach outperforms existing SSMIS methods and is comparable to fully supervised methods. Code has been released at GitHub

Super-resolution plays a crucial role in enhancing the diagnostic utility of ultrasound images by recovering fine anatomical structures. Due to their remarkable performance in natural image restoration and generation, diffusion probabilistic models have become attractive for ultrasound image superresolution that highly requires complex textures and structural details modelling. However, existing diffusion-based superresolution methods typically apply uniform noise perturbations and reconstruction strategies across the entire image. Such global treatment neglects the varying reconstruction difficulty and clinical importance of different regions, often resulting in blurry lesion boundaries and suboptimal recovery of diagnostically significant structures. To address this issue, we propose UGDT-SR, a structure-aware ultrasound image super-resolution framework that integrates Bayesian uncertainty modeling and Transformer-guided structural representation into the diffusion process. Specifically, we introduce a Bayesian uncertainty network to estimate pixel-wise reconstruction difficulty, generating structural uncertainty masks that adaptively modulate local noise injection during forward diffusion. Furthermore, we inject these uncertainty masks into the conditional encoder to guide the model's attention toward clinically relevant regions throughout the reconstruction. Extensive experiments on both public and clinical ultrasound datasets demonstrate that UGDT-SR consistently outperformed CNN-, GAN-, and diffusion-based baselines, achieving superior perceptual quality and structural fidelity, especially in lesion areas.

3D Gaussian Splatting (3D-GS) is a recent 3D scene reconstruction technique that enables real-time rendering of novel views by modeling scenes as parametric point clouds of differentiable 3D Gaussians. However, its rendering speed and model size still present bottlenecks, especially in resource-constrained settings. In this paper, we identify and address two key inefficiencies in 3D-GS to substantially improve rendering speed. These improvements also yield the ancillary benefits of reduced model size and training time. First, we optimize the rendering pipeline to precisely localize Gaussians in the scene, boosting rendering speed without altering visual fidelity. Second, we introduce a novel pruning technique and integrate it into the training pipeline, significantly reducing model size and training time while further raising rendering speed. Our Speedy-Splat approach combines these techniques to accelerate average rendering speed by a drastic 6.71× across scenes from the Mip-NeRF 360, Tanks & Temples, and Deep Blending datasets. Code and additional visualizations are available at https://speedysplat.github.io/.

Weakly-supervised learning methods have become increasingly attractive for medical image segmentation, but suffered from a high dependence on quantifying the pixel-wise affinities of low-level features, which are easily corrupted in thyroid ultrasound images, resulting in segmentation over-fitting to weakly annotated regions without precise delineation of target boundaries. We propose a dual-branch weakly-supervised learning framework to optimize the backbone segmentation network by calibrating semantic features into rational spatial distribution under the indirect, coarse guidance of the bounding box mask. Specifically, in the spatial arrangement consistency branch, the maximum activations sampled from the preliminary segmentation prediction and the bounding box mask along the horizontal and vertical dimensions are compared to measure the rationality of the approximate target localization. In the hierarchical prediction consistency branch, the target and background prototypes are encapsulated from the semantic features under the combined guidance of the preliminary segmentation prediction and the bounding box mask. The secondary segmentation prediction induced from the prototypes is compared with the preliminary prediction to quantify the rationality of the elaborated target and background semantic feature perception. Experiments on three thyroid datasets illustrate that our model outperforms existing weakly-supervised methods for thyroid gland and nodule segmentation and is comparable to the performance of fully-supervised methods with reduced annotation time. The proposed method has provided a weakly-supervised segmentation strategy by simultaneously considering the target's location and the rationality of target and background semantic features distribution. It can improve the applicability of deep learning based segmentation in the clinical practice.

Learning probabilistic models that can estimate the density of a given set of samples, and generate samples from that density, is one of the fundamental challenges in unsupervised machine learning. We introduce a new generative model based on denoising density estimators (DDEs), which are scalar functions parametrized by neural networks, that are efficiently trained to represent kernel density estimators of the data. Leveraging DDEs, our main contribution is a novel technique to obtain generative models by minimizing the Kullback-Leibler (KL)-divergence directly. We prove that our algorithm for obtaining generative models is guaranteed to converge consistently to the correct solution. Our approach does not require specific network architecture as in normalizing flows (NFs), nor use ordinary differential equation (ODE) solvers as in continuous NFs. Experimental results demonstrate substantial improvement in density estimation and competitive performance in generative model training.

Video matting has broad applications, from adding interesting effects to casually captured movies to assisting video production professionals. Matting with associated effects such as shadows and reflections has also attracted increasing research activity, and methods like Omnimatte have been proposed to separate dynamic foreground objects of interest into their own layers. However, prior works represent video backgrounds as 2D image layers, limiting their capacity to express more complicated scenes, thus hindering application to real-world videos. In this paper, we propose a novel video matting method, OmnimatteRF, that combines dynamic 2D foreground layers and a 3D background model. The 2D layers preserve the details of the subjects, while the 3D background robustly reconstructs scenes in real-world videos. Extensive experiments demonstrate that our method reconstructs scenes with better quality on various videos.

Abstract Witnessing broad energy‐critical applications of barrier coverage in mobile and wireless sensor networks, emerging practical applications have recently brought a new barrier coverage model which uses sink‐based mobile sensors for covering a given barrier with the aim of prolonging the lifespan of the coverage. In the model, a set of sink stations were distributed on the plane in which each sink can emit mobile sensors with an identical radius. The task is to cover a given line barrier with the emitted mobile sensors, aiming to minimize the maximum movement of the sensors so as to prolong the shortest lifespan among the sensors in coverage. In this paper, we first devise an algorithm for optimally solving the problem based on the properties of the structures called movement parity and tangent equilibrium points between the sinks. Then based on a more sophisticated geometric property of optimum solutions, we improve the runtime to a linear runtime O(k) which attains the possibly optimum runtime of the problem for k being the number of sinks. At last, numerical experiments are carried out to demonstrate the practical performance gain of our algorithms against baselines in literature.

3D reconstruction from images is a core problem in computer vision. With recent advances in deep learning, it has become possible to recover plausible 3D shapes even from single RGB images. However, obtaining detailed geometry and texture for objects with arbitrary topology remains challenging. In this paper, we propose a novel approach for reconstructing point clouds from RGB images. Unlike other methods, we can recover dense point clouds with hundreds of thousands of points, and we also include RGB textures. In addition, we train our model on multiple categories, which leads to superior generalization to unseen categories com-pared to previous techniques. We achieve this using a two-stage approach, where we first infer an object coordinate map from the input RGB image, and then obtain the final point cloud using a reprojection and completion step. We show results on standard benchmarks that demonstrate the advantages of our technique.

As a fundamental and critical task in feature-based remote sensing image registration, feature matching refers to establishing reliable point correspondences from two images of the same scene. In this article, we propose a simple yet efficient method termed linear adaptive filtering (LAF) for both rigid and nonrigid feature matching of remote sensing images and apply it to the image registration task. Our algorithm starts with establishing putative feature correspondences based on local descriptors and then focuses on removing outliers using geometrical consistency priori together with filtering and denoising theory. Specifically, we first grid the correspondence space into several nonoverlapping cells and calculate a typical motion vector for each one. Subsequently, we remove false matches by checking the consistency between each putative match and the typical motion vector in the corresponding cell, which is achieved by a Gaussian kernel convolution operation. By refining the typical motion vector in an iterative manner, we further introduce a progressive strategy based on the coarse-to-fine theory to promote the matching accuracy gradually. In addition, an adaptive parameter setting strategy and posterior probability estimation based on the expectation-maximization algorithm enhance the robustness of our method to different data. Most importantly, our method is quite efficient where the gridding strategy enables it to achieve linear time complexity. Consequently, some sparse point-based tasks may inspire from our method when they are achieved by deep learning techniques. Extensive feature matching and image registration experiments on several remote sensing data sets demonstrate the superiority of our approach over the state of the art.

In this paper, we explore to learn representations of legislation and legislator for the prediction of roll call results. The most popular approach for this topic is named the ideal point model that relies on historical voting information for representation learning of legislators. It largely ignores the context information of the legislative data. We, therefore, propose to incorporate context information to learn dense representations for both legislators and legislation. For legislators, we incorporate relations among them via graph convolutional neural networks (GCN) for their representation learning. For legislation, we utilize its narrative description via recurrent neural networks (RNN) for representation learning. In order to align two kinds of representations in the same vector space, we introduce a triplet loss for the joint training. Experimental results on a self-constructed dataset show the effectiveness of our model for roll call results prediction compared to some state-of-the-art baselines.

Energy Internet (EI) is aimed at sustainable computing by integrating various energy forms into a highly flexible grid similar to the Internet. The network subsystems of EI connect different components to enable real-time monitoring, controlling, and management. In this paper, the spectrum allocation problem of the cognitive radio network for EI in a smart city is investigated. The network spectrum allocation with both heterogeneous primary operators and secondary users is formulated as the combinatorial auction problem and then is converted to a subset selection problem on a weighted bipartite graph. We propose a hybrid algorithm to solve the problem. Firstly, the proposed algorithm uses a constructive procedure based on the Kuhn-Munkres algorithm to obtain an initial solution. Then, a local search is used to improve the solution quality. In addition, the truthfulness of the auction is guaranteed by adopting a “Vickrey-like” mechanism. Simulation results show that the performance of the proposed algorithm is better than existing greedy algorithms in terms of the social welfare, seller revenue, buyer satisfaction ratio, and winning buyer ratio.

The minimum weight-dominating set (MWDS) problem is NP-hard and has a lot of applications in the real world. Several metaheuristic methods have been developed for solving the problem effectively, but suffering from high CPU time on large-scale instances. In this paper, we design an effective hybrid memetic algorithm (HMA) for the MWDS problem. First, the MWDS problem is formulated as a constrained 0-1 programming problem and is converted to an equivalent unconstrained 0-1 problem using an adaptive penalty function. Then, we develop a memetic algorithm for the resulting problem, which contains a greedy randomized adaptive construction procedure, a tabu local search procedure, a crossover operator, a population-updating method, and a path-relinking procedure. These strategies make a good tradeoff between intensification and diversification. A number of experiments were carried out on three types of instances from the literature. Compared with existing algorithms, HMA is able to find high-quality solutions in much less CPU time. Specifically, HMA is at least six times faster than existing algorithms on the tested instances. With increasing instance size, the CPU time required by HMA increases much more slowly than required by existing algorithms.

The max-bisection problem consists in partitioning the vertices of a weighted undirected graph into two equally sized subsets so as to maximize the sum of the weights of crossing edges. It is an NP-hard combinatorial optimization problem that arises in many applications. In this paper, we present a memetic algorithm for the max-bisection problem, which integrates a new fast local search procedure, a crossover operator, and a pool updating strategy. These strategies achieve a balance between intensification and diversification. Extensive experiments were performed on a number of benchmark instances with 800 to 10,000 vertices from the literature. The proposed memetic algorithm improved the best known solutions for all benchmark instances tested in this paper. The improvement in terms of cut value over the CirCut by Burer et al.ranging from 0.02 to 4.15 percent, and the average time of our proposed memetic algorithm is much lower than that of CirCut. It shows that the proposed memetic algorithm can find high quality solutions in an acceptable running time.

Software-defined networking (SDN) is envisioned to provide a centralized interface to programmatically manage networking elements. However, despite its conceptual simplicity, current switch and SDN architectures have poor performance with little support to innovate and test novel SDN applications. We propose an application extensibility framework that allows researchers to build new SDN applications without requiring modification to the OpenFlow-based plumbing available today. We also employ both hardware and software packet processing capabilities of switching elements by offloading intensive per-packet processing onto the switch processing pipeline using dynamically loadable packet processing modules (PPMs). Our architecture thus allows flexibility in the type of applications alongside high switching performance. We believe that our architecture will unleash the potential of SDN by inspiring the SDN “killer app”. We evaluate our framework using an encryption middlebox application and show a two orders-of-magnitude improvement over an implementation using the current SDN architecture when using hardware offload blocks.

In this paper, we propose a “multistart-type” algorithm for solving the max-k-cut problem. Central to our algorithm is an auxiliary function we propose. We formulate the max-k-cut problem as an explicit mathematical form, which allows us to use an easy implementable local search. The construction of the auxiliary function requires a local maximizer of the max-k-cut problem. If the best local maximizer obtained is used in the construction of the auxiliary function, then the local maximization of the auxiliary function leads to a better maximizer of the max-k-cut problem. This proves to be a good strategy to escape from the current local optima and to search a broader solution space. Indeed, we have shown, both numerically and theoretically, that the maximization of the auxiliary function by the local search method can escape successfully from previously converged discrete local maximizers by taking increasing values of a parameter. Computational results on many test instances with different sizes and densities show that the proposed algorithm is efficient and stable to find approximate global solutions for the max-k-cut problems. Although we have presented results for k ≥ 2, the robustness of our algorithm is shown for k = 2 by comparisons with a number of recent methods. A number of theoretical results are also presented, which justify the design of our algorithm.

This panel brings together technology experts and business leaders and provide first hand insight to the evolution of cloud computing and IT as a service (ITaaS), from both technology and business model perspectives. The panel will discuss the disruptive nature of cloud computing and its business model, including the impact to the current enterprise IT industry, to the service provider industry, to the enterprise software industry, to the networking industry, and to the service industry. The panel will also discuss the confluence of SOA paradigm and SaaS paradigm and examine its implication to the enterprise IT architecture. The panel will also help audience understand the limitation and challenges for cloud computing and ITaaS. The audience of this panel is targeted at the technology leaders and business decision makers in enterprise IT, software industry, and networking industry.

This paper presents a service on demand management framework to support dynamic service creation, and a system which instantiates this framework for an Ethernet-To-The-Home/Business (ETTx) service environment. The service-on-demand framework delivers two major functions: (1) it allows service providers to create new types of service dynamically through a policy-driven "service building" process; (2) it allows end users to subscribe to a service and have their service orders fulfilled in a fully automated, near real time, and "zero touch" fashion. We have applied this framework to build an industrial strength service fulfillment system geared towards ETTx access networks. Those networks represent one of the most dynamic service creation environments today.

Summary form only given. This tutorial presents a view of the IP network management problem from the perspective of the network itself. It focuses on how to address enhanced manageability requirements in the network infrastructure. This includes what can and should be instrumented inside a network element, and what can and should be instrumented with a scope that goes beyond the individual network element to result in more general "manageability services" provided by the network. We discuss the relevance of manageability as a business driver and provide an in-depth analysis of the various factors that contribute to manageability. After this, techniques are presented that can be applied to enhance manageability, also in light of increased focus on applications such as on-demand services, managed services, and plug-n-play networking. We also discuss the role of certain technologies such as XML and Netflow in the emergence of a new generation of management interfaces. Finally, scenarios are used to illustrate how the resulting advances in manageability facilitate not only the development of management applications, but ultimately result in networks that are more autonomous and "self managing", and less complex and costly to operate than the norm today.

Switching networks with probabilistic traffic are positioned prominently in communication engineering. Measures of performance for such a network include the blocking probability of the network and the time for the routing algorithm to establish communication paths. Although literature exists concerning blocking probability, little theoretical progress in efficient routing algorithms has been made. Since the network is under a probabilistic traffic, it is meaningful to measure the routing algorithm by its expected running time. In this paper, we consider routing algorithms for a class of networks known as series-parallel networks. We first prove a lower bound for the expected time for any routing algorithm to establish a communication path and then present an algorithm that has an expected time within a constant factor of the lower bound, thus establishing the optimality of the algorithm. © 1996 John Wiley & Sons, Inc.

The authors consider fault-tolerant circuit-switching networks under a random switch failure model. Three circuit-switching networks of theoretical importance—nonblocking networks, rearrangeable networks, and superconcentrators—are studied. The authors prove lower bounds for the size (the number of switches) and depth (the largest number of switches on a communication path) of such fault-tolerant networks and explicitly construct such networks with optimal size $\Theta ( n ( \log n )^2 )$ and depth $\Theta ( \log n )$.

Circuit-switching networks are used to support simultaneous (data, voice and image) communications across multiprocessor parallel systems, distributed computer systems, and telecommunication systems.Circuit-switching networks accomplish simultaneous communications by means of disjoint paths of electrical links and switches.In this paper, we shall consider fault-tolerant circuit-switching networks under a random switch failure model.Three circuit-switching net works of theoretical importance, non-blocking networks, rearrange able networks and superconcent raters, are studied.We prove lower bounds for the size (the number of switches) and depth (the largest number of switches on a communication path) of such faulttolerant networks.And we explicitly construct such networks with optimal sizes and depths.

Article Fault tolerant planar communication networks Share on Author: Geng Lin Department of Computer Science, The University of British Columbia, Vancouver, British Columbia V6T 1Z2, Canada Department of Computer Science, The University of British Columbia, Vancouver, British Columbia V6T 1Z2, CanadaView Profile Authors Info & Claims STOC '92: Proceedings of the twenty-fourth annual ACM symposium on Theory of ComputingJuly 1992 Pages 133–139https://doi.org/10.1145/129712.129726Online:01 July 1992Publication History 3citation173DownloadsMetricsTotal Citations3Total Downloads173Last 12 Months1Last 6 weeks0 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my AlertsNew Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteGet Access

Non-blocking networks have many applications in communications.Typical examples are telephone switching networks and communication networks among processors or between processors and memory