Sliced inverse regression (SIR) is an innovative and effective method for sufficient dimension reduction data visualization. Recently, impressive range of penalized SIR methods has been proposed to estimate the central subspace in a sparse fashion. Nonetheless, few them considered from decision-theoretic point view. To address this issue, we paper establish minimax rates convergence estimating directions under various commonly used loss functions literature reduction. We also discover...

Hyperspectral imaging (HSI) systems can provide sensitive and specific detection identification of high value targets in the presence complex backgrounds. However, current generation sensors are typically large costly to field, do not usually operate real-time. Sensors that capable real-time operation have compromise on number spectral bands, image definition, and/or being detected. Additionally, these command a cost designed configured for mission profiles, making them unable adapt multiple...

Ultra-limited-angle image reconstruction problem with a limited-angle scanning range less than or equal to π2 is severely ill-posed. Due the considerably large condition number of linear system for reconstruction, it extremely challenging generate valid reconstructed by traditional iterative algorithms.To develop and test ultra-limited-angle CT algorithm.We propose new optimized model Reweighted Alternating Edge-preserving Diffusion Smoothing algorithm in which reweighted method improving...

This article researches on node importance ranking of multiplex network. Existing methods are amost monoplex network using only single relationship. However, in real life a person often has various relationships, so how to rank nodes is more realistic and meaningful. In this article, we propose method based weighted aggregation. Firstly, weight each layer by Analytic Hierarchy Process for existing aggregation either lack weighting process or rely artificial calculation rather than precise...

In compressed sensing, a small enough restricted isometry constant (RIC) of the sensing matrix satisfying property (RIP) is powerful guarantee on precise reconstruction sparse discrete signal. Under certain condition, RIC can be improved by weighting so that its all nonzero singular values become same, i.e., condition number new 1. this article, multiplying linear system A <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</sup> Ax = y related to...

BACKGROUND: The limited-angle reconstruction problem is of both theoretical and practical importance. Due to the severe ill-posedness problem, it very challenging get a valid reconstructed result from known small projection data. leads normal equation AT Ax = b linear system derived by discretizing Radon transform be severely ill-posed, which quantified as large condition number A. OBJECTIVE: To develop test new algorithm for improving image with appropriately angle range [ 0 , π 3 ] ∼ 2 ....

When a measurement falls outside the quantization or measurable range, it becomes saturated and cannot be used in classical reconstruction methods. For example, C-arm angiography systems, which provide projection radiography, fluoroscopy, digital subtraction angiography, are widely for medical diagnoses interventions, limited dynamic range of flat detectors leads to overexposure some projections during an acquisition, such as imaging relatively thin body parts (e.g., knee). Aiming at...

In compressed sensing (CS), coherence is a simple and practical measure of the quality matrix. The smaller matrix, better reconstruction result. Most strategies for optimizing matrix are only applicable to gray-value matrices not binary with fast computation speed storage space. Under certain condition, can be improved by weighting make its condition number equal 1, i.e., all non-zero singular values become same. this paper, we propose method reweighting CS models reduce so as improve...

the public expects to obtain travel experience with low cost, convenient and high comfort, but also has different tour needs such as history, culture, natural landscape food shopping. Therefore, this paper presents a personalized route recommendation method based on modified Shuffled Frog Leaping Algorithm. First, optimization model of is established, Algorithm designed improved according characteristics model. By adjusting controllable accuracy, increasing screening criteria enhancing...

The robust principal component analysis (RPCA) decomposes a data matrix into low-rank part and sparse part. There are mainly two types of algorithms for RPCA. first type algorithm applies regularization terms on the singular values to obtain matrix. However, calculating can be very expensive large matrices. second replaces as multiplication small They faster than because no value decomposition (SVD) is required. rank required, an accurate estimation needed reasonable solution. In this paper,...

The one-bit quantization is implemented by one single comparator that operates at low power and a high rate. Hence compressive sensing (1bit-CS) becomes attractive in signal processing. When measurements are corrupted noise during acquisition transmission, 1bit-CS usually modeled as minimizing loss function with sparsity constraint. one-sided $\ell_1$ the linear two popular functions for 1bit-CS. To improve decoding performance on noisy data, we consider pinball loss, which provides bridge...

In recent years, there have been significant advancements in deep learning-based lightweight image super-resolution (SR) reconstruction techniques. However, practical applications, are still challenges to be addressed. Many existing SR algorithms simplify the model by reducing number of parameters or changing combination convolutions, which results significantly decreased performance and stability, leading poor results. To address this issue, we proposed a Cosine Self-Attention mechanism...

Video summarization can help people retrieve videos quickly. Existing unsupervised video models make insufficient use of frame uniqueness in the self-attention mechanism and have redundancy computation uniqueness. In this paper, we rethink relationship between attentive diversity frames, improve calculation method by masking adjacent information model framework based on a mechanism, simplify concentrated attention with it. addition, propose new loss function called Deviation loss, which uses...

Effective segment extraction has broad application prospects in the field of video surveillance. In action recognition or object recognition, a large number repetitive background segments will reduce efficiency computation. This causes algorithm to consume more time. can greatly save computing and storage resources for both target detection storage. Common algorithms are process whole surveillance video, such as frame difference method optical flow method. However, false occur when picture...

Abstract We consider approximation of multivariate functions in Sobolev spaces by high order Parzen windows a non-uniform sampling setting. Sampling points are neither i.i.d. nor regular, but noised from regular grids shifts probability density function. Sample function values at drawn according to measures with expected being the approximated The orders estimated means regularity function, and windows, under suitable choices scaling parameter.

Karyotyping requires chromosome instances to be segmented and classified from the metaphase images. One of difficulties in segmentation is that chromosomes are randomly positioned image, there a great chance for touched or overlap with others. It always much easier operators automatic programs tackle images without overlapping than ones largely overlapped chromosomes. In order reduce processing difficulty, adding smart image selection procedure ahead practical necessary. this paper, we...