The distributed nonconvex optimization problem of minimizing a global cost function formed by sum <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$n$</tex> local functions using information exchange is considered. This an important component many machine learning techniques with data parallelism, such as deep and federated learning. We propose primal-dual stochastic gradient descent (SGD) algorithm, suitable for arbitrarily connected...

In modern industry, human–robot collaboration is becoming the norm. Since robots need to share same workspace with humans in an unstructured/semistructured environment, robot–human and robot–environment collisions are inevitable general. To reduce harm caused by these collisions, it necessary detect them real time so that actions can be taken accordingly. this article, we propose a general robot collision detection method based on switched momentum dynamics identification. This enables...

The authors analyse the contour error dynamics in contouring control of dual‐arm robotic manipulators with holonomic constraints. With modified robot, robot constraints is symetrically transformed into problem by considering equivalent method. It becomes stabilisation such that it suitable for robust approach to improve performance terms control, i.e. accuracy high speed. proposed method can deal both analytic and non‐analytic functions desired path task space. experimental results reveal...

In this paper, we present a unified framework based on integral quadratic constraints for analyzing the convergence of distributed push-pull optimization algorithms directed graphs. Our provides numerical upper bounds linear rates existing when local objective functions are strongly convex and smooth graphs connected. Moreover, propose new algorithm show that proposed can also be applied to establish its rate. The theoretical results illustrated validated via examples.

This paper considers the distributed optimization problem of minimizing a global cost function formed by sum local smooth functions using information exchange. A standard assumption for proving exponential/linear convergence existing first-order methods is strong convexity functions. does not hold many practical applications. In this paper, we propose continuous-time primal-dual gradient descent algorithm and show that it converges exponentially to minimizer under satisfies restricted secant...

Through the development of multi-modal and contrastive learning, image video retrieval have made immense progress over last years. Organically fused text, image, knowledge brings huge potential opportunities for multi-dimension, multi-view retrieval, especially in traffic senses. This paper proposes a novel Multimodal Language Vehicle Retrieval (MLVR) system, retrieving trajectory tracked vehicles based on natural language descriptions. The MLVR system is mainly combined with an end-to-end...

The distributed nonconvex optimization problem of minimizing a global cost function formed by sum $n$ local functions using information exchange is considered. This an important component many machine learning techniques with data parallelism, such as deep and federated learning. We propose primal--dual stochastic gradient descent (SGD) algorithm, suitable for arbitrarily connected communication networks any smooth (possibly nonconvex) functions. show that the proposed algorithm achieves...

This paper investigates how to accelerate the convergence of distributed optimization algorithms on non-convex problems with only zeroth-order information available. We propose a (ZO) primal-dual stochastic coordinate algorithm equipped "powerball" method convergence. establish result considered for general cost functions. verify proposed through benchmark example generating adversarial examples from black-box DNNs in order compare existing state-of-the-art centralized and ZO algorithms. The...

This paper considers the distributed nonconvex optimization problem of minimizing a global cost function formed by sum local functions using information exchange. We first consider first-order primal-dual algorithm. show that it converges sublinearly to stationary point if each is smooth and linearly optimum under an additional condition satisfies Polyak-{\L}ojasiewicz condition. weaker than strong convexity, which standard for proving linear convergence algorithms, minimizer not necessarily...

Using the method of adaptive kernel learning based relevance vector machine (ARVM) and combining morphological filtering clustering criterion recommended by Kallergi, a new algorithm for microcalcification (MC) clusters processing in mammograms is investigated. Firstly, detection MC formulated as supervised-learning problem. Then ARVM used classifier to determine whether an object present at each location mammogram remove isolated spurious pixels. Finally, identified are obtained Kallergi...

We show an original method for automatic hand gesture recognition that makes use of fuzzified latent-dynamic conditional random fields (LDCRF). In this method, fuzzy linguistic variables are used to model the features gestures and then modify potential function in LDCRFs. By combining LDCRFs sets, these fuzzy-based (FLDCRF) have advantages sequence labeling along with advantage retaining imprecise character gestures. The efficiency proposed was tested unsegmented sequences three different...

Data outsourcing has become quite popular in recent years. While the rapid development of cloud storage service, it brings about many new security challenges. One biggest concerns with data is that integrity at untrusted servers. For example, when individual users store important files servers, may be damaged or corrupted because server failures malicious attackers. Our work focuses on problem ensuring computing. The prior approaches mainly solved static and provide simple recovery method...

Effective segmentation of microcalcifications in mammograms is crucial for the quantification morphologic properties by features incorporated computer-aided diagnosis schemes. A multi-resolution region growth method based on edge feature proposed this paper, which vectors are used to obtain complete microcalcifications. Then a microcalcification and image difference presented. By using method, under precondition high detecting rate, can be segmented. What more, shape distribution obtained...

This paper considers the distributed smooth optimization problem in which objective is to minimize a global cost function formed by sum of local functions, using information exchange. The standard assumption for proving exponential/linear convergence first-order methods strong convexity does not hold many practical applications. In this paper, we first show that continuous-time primal-dual gradient algorithm converges one minimizer exponentially under satisfies restricted secant inequality...

Abstract With the development and wide application of internet, cloud computing technology with powerful functions has been initially applied in fields industry, agriculture, scientific research, medical treatment so on. Its technical advantages undoubtedly bring power to library. Intelligent library is advanced form intelligent Cloud service makes user’s data storage software operation internet server, which network high integration, efficient. On basis discussing basic meaning computing,...

In recent years, there has been rapid development in 3D generation models, opening up new possibilities for applications such as simulating the dynamic movements of objects and customizing their behaviors. However, current generative models tend to focus only on surface features color shape, neglecting inherent physical properties that govern behavior real world. To accurately simulate physics-aligned dynamics, it is essential predict materials incorporate them into prediction process....