Survival analysis is extensively employed to analyze the probability of the event of interest, particularly in the medical field. Most current research treats patients as isolated entities, neglecting the complex associations among them, which leads to underutilization of valuable information. Recently, several studies address this limitation by incorporating patient graph structures. However, these approaches generally overlook two critical issues: 1) the exploration of heterogeneous inter-patient relationships, and 2) flexible and scalable inductive inference for test samples. To overcome these challenges, this study introduces a novel framework, Multiplex Graph Guided Deep Survival Analysis (MGG-Surv). Specifically, we employ multiplex patient graphs to capture comprehensive inter-patient associative information. Furthermore, we propose a teacher-student dual network architecture, where the teacher network encodes multiplex graphs, and the learned graph knowledge is transferred to the student network via a unidirectional connection termed Graph-Guided Distillation. The student network integrates this graph knowledge to predict survival outcomes without requiring the patient graphs. These innovative designs facilitate comprehensive integration of inter-patient relationships while achieving flexible and scalable graph-free inference. Experiments on four datasets, encompass-ing both single and competing risks, demonstrate the superior performance of our framework.

Immune thrombocytopenia (ITP) is a typically self-limiting and immune-mediated bleeding disorder in children. Approximately 20% of children with ITP experience chronicity, leading to reduced quality of life and increased treatment burden. The accurate prediction of chronicity would enable clinicians to make personalized treatment plans at an early stage. However, due to the self-limiting nature of ITP and the scarcity of available children patients, the data presents two prominent issues: small data and imbalanced class, which are unfavorable for effectively training a deep learning model. To handle these issues concurrently, we proposed a novel method that integrates contrastive learning with the Transformer. First, we adopt the FT-Transformer as our backbone, which allows our model to flexibly process heterogeneous tabular data. Second, we amplify and balance the original data via random masking and oversampling, respectively. Lastly, we build contrastive pairs according to the latent representations generated by the FT-Transformer encoder, such that the amplified and oversampled synthetic data can be utilized thoroughly. The experimental results on real-world ITP children data show that our proposal outperforms the state-of-the-art methods, and demonstrate the significant advantages of dealing with insufficient and imbalanced problems.

Survival analysis is employed to analyze the time before the event of interest occurs, which is broadly applied in many fields. The existence of censored data with incomplete supervision information about survival outcomes is one key challenge in survival analysis tasks. Although some progress has been made on this issue recently, the present methods generally treat the instances as separate ones while ignoring their potential correlations, thus rendering unsatisfactory performance. In this study, we propose a novel Deep Survival Analysis model with latent Clustering and Contrastive learning (DSACC). Specifically, we jointly optimize representation learning, latent clustering and survival prediction in a unified framework. In this way, the clusters distribution structure in latent representation space is revealed, and meanwhile the structure of the clusters is well incorporated to improve the ability of survival prediction. Besides, by virtue of the learned clusters, we further propose a contrastive loss function, where the uncensored data in each cluster are set as anchors, and the censored data are treated as positive/negative sample pairs according to whether they belong to the same cluster or not. This design enables the censored data to make full use of the supervision information of the uncensored samples. Through extensive experiments on four popular clinical datasets, we demonstrate that our proposed DSACC achieves advanced performance in terms of both C-index (0.6722, 0.6793, 0.6350, and 0.7943) and Integrated Brier Score (IBS) (0.1616, 0.1826, 0.2028, and 0.1120).

Federated learning (FL) faces many security threats. Although multiple robust FL frameworks have been proposed to defend against these malicious attacks in horizontal federated learning (HFL), security issues in vertical federated learning (VFL) have not been adequately studied. Recent studies show that VFL is vulnerable to inference attacks (e.g., label inference attacks), which puts VFL at risk. To solve this problem, we propose a new VFL framework SVFL (Secure Vertical Federated Learning) to defend against privacy breaches inspired by feature disentanglement. Specifically, in SVFL , the bottom models are feature extractors to extract samples’ features in the high-dimensional space, and the top model sews samples’ features of the same sample ID. Then, disentangling the samples’ features into the class-relevant feature and class-irrelevant one via two classifiers: one is to recognize the class-relevant feature by regular training, and another is to recognize the class-irrelevant feature by adversarial training . Our experiments show that SVFL not only defends against label inference attacks, no matter how many samples features a malicious participant occupies, but also improves the global model’s accuracy. Therefore, SVFL provides a privacy security guarantee for the vertical federated learning system .

A GEOsynchronous Spaceborne-Airborne Bistatic Synthetic Aperture Radar (GEO SA-BSAR) system has been proved to be a significant tool for moving targets monitoring. Due to the special geometry model of the GEO SA-BSAR system, there is a complex relative movement between the moving target and the bistatic radar, leading to an additional phase modulation of the echo and further, causing moving targets to be smeared in the SAR image. Recently, Deep Neural Network (DNN) shows great potential in rapid image recovery. However, most image recovery methods based on DNN concentrate on the whole image, which limits the imaging performance of sparse targets. In this letter, we propose a DNN framework with similarity constraints for GEO SA-BSAR moving target imaging. This DNN-based method optimizes the cosine similarity of azimuth signals between the ground-truth image and the predicted image in the loss function to recover the azimuth position and focusing characteristics of the sparse targets. Extensive experimental results prove that the proposed model can quickly obtain GEO SA-BSAR moving target images with small training datasets compared with some counterparts.

A long aperture time is required to achieve a high signal-to-noise ratio and high azimuth resolution in geosynchronous spaceborne-airborne bistatic synthetic aperture radar (GEO SA-BSAR) system for moving target indication (MTI). The range walk migration because of the target's motion cannot be ignored for such a long time, and the second-order range model fails for the moving target. In this paper, an adaptive MTI method is proposed for GEO SA-BSAR with a long aperture time, ensuring the moving target's detection and location. Firstly, an adaptive spatial filter modified by the accurate GEO SA-BSAR multichannel signal model is applied to clutter suppression and beamforming. Next, we adopt the generalized Radon-Fourier transform to maximize the signal-to-noise ratio of the moving target with unknown motion parameters. Then, the moving target can be detected, and its position has been obtained. Finally, the simulation experiments are conducted to show the effectiveness of our technique.

Geosynchronous spaceborne-airborne bistatic synthetic aperture radar (GEO SA-BSAR), consisting of GEO transmitter and airborne receiver, has stable coverage for a long time and benefits moving target detection. The performance of GEO SA-BSAR moving target indication (GEO SA-BSAR MTI) system is investigated. This work provides the first analytical connection between target detection performance and configuration parameters based on the optimum output signal to clutter and noise ratio (SCNR) criterion. Furthermore, a theoretical analysis of motion parameter estimation is stated in terms of the Cramer-Rao lower bounds (CRLB). The SCNR loss and CRLB provides a necessary design configuration for GEO SA-BSAR MTI system. Finally, the optimized configuration is provided by comparing the performance.

Multistatic geosynchronous synthetic aperture radar (GEO SAR) utilizes multiple satellites' transmitted and received signals simultaneously to generate multiple phase centers (PCs) and to reduce the synthetic aperture time and the power budget. Two key problems of this SAR system are the design of the PC array configuration to satisfy the resolution requirements and the suppression of the grating lobes that are inherently introduced by the PC array. The resolution of the SAR can be analyzed using generalized ambiguity functions (GAFs). However, the current research on the multistatic SAR resolution and grating lobes is mostly based on specified configurations and numerical simulation methods, which cannot establish the relationship between the system parameters and the resolution. This article proposes an analytical analysis method for the multistatic GEO SAR GAF-based on the array spatial ambiguity function (ASAF). First, gradient analysis is used to obtain analytical expressions for ASAF and the multistatic SAR GAF. On this basis, an analytical resolution expression is obtained, and an orbital element design method that considers the Earth's rotation is proposed. In addition, the lobe positions are analytically expressed based on the geometry, and the grating lobes are suppressed by designing the optimal integration time such that the null depth of the velocity ambiguity function (VAF) coincides with the grating lobe positions. Finally, simulation results at various positions for various orbit types demonstrate the accuracy of the GAF approximation formula and the satisfactory performance of the optimal time expression in suppressing the grating lobes.

Despite the increasing interest in geosynchronous synthetic aperture radar (GEO SAR) systems, the ground moving target indication and motion parameters estimation aspect have never been addressed in GEO SAR scenarios. In this article, we tackle the issue of multiple moving target velocity estimation in GEO SAR. We develop new closed-form expressions that relate both Doppler centroid and Doppler rate to the target motion parameters in GEO SAR by considering the specific features of a geosynchronous orbit. Furthermore, we propose a new velocity estimation algorithm that combines the nonuniform cubic phase function (NU-CPF) algorithm with the newly developed models to estimate the moving target's two velocity components. Moreover, based on the above-mentioned technique, we further propose a solution to address the multiple moving targets' problem. Simulation results along with an estimation accuracy analysis are provided to demonstrate the effectiveness of the proposed multitarget GEO SAR velocity estimation technique.

In color image compression, the chroma components are often sub-sampled before compression and up-sampled after compression. Although sub-sampling the chroma components saves the bit-cost for compression, it often induces extra color distortions in the compressed images. In this paper, we propose two approaches to tackle this problem. First, we propose a sub-sampling method in the transform domain and apply it to both chroma components. Then, based on this sub-sampling, we propose a novel method to modify the luma component. In our proposed luma modification algorithm, the distortions that occurred in the two chroma components can be coupled together and utilized to modify the luma component. With our proposed chroma sub-sampling and luma modification algorithms, we can achieve a low RGB distortion in practical image coding. The experimental results demonstrate that our proposed methods offer more significant coding gains compared with the state-of-the-art methods for the compression of color images.

Transform domain down-scaling (TDDS) is traditionally implemented by dropping most of high-frequency components of the transformed block. Applying it to image compression can improve the compression efficiency by saving considerable bit-cost. Due to losing some necessary high-frequency information, the resulted image compressed by using the traditional TDDS-based coding often suffers a serious quality degradation. In this paper, we propose a compression-constrained TDDS and perform it on each N × N block to produce an N/2 × N/2 coefficient block for the compression. Our proposed TDDS not only guarantees a high reconstruction quality but also makes a low bit-cost for compression. We integrate it in practical image coding to build up our proposed compression scheme. Experimental results show that our proposed method demonstrates excellent coding performance when used to compress image signals.

Geosynchronous SAR (GEO SAR) has better coverage and revisiting performance than the low Earth orbit SAR. Thus GEO SAR can implement the repeat-pass interferometry as its revisit time of 1 day and the orbit drift (forming baseline) induced by perturbation. In comparison, the twin GEO SAR formation configuration in which the slave satellite repeats the master satellite' track can realize shorter revisit time and adjustable baseline by tuning the orbit elements. In this paper, a validation experiment based on the Chinese Beidou IGSO navigation satellites is demonstrated to validate the feasibility and effectiveness of this concept. The validation experiment consists of data acquisition of the Beidou IGSO satellites signals, equivalent pre-processing, image processing and interferometry processing. In the end, we obtain the high-precision experimental results of deformation inversion, validating the feasibility of twin GEO SAR interferometry/differential interferometry.

The 3-D deformation retrieval is significant for the accurate evaluation of geologic disasters (e.g., earthquakes and landslides). Multiple-aperture interferometry (MAI) is an effective method to obtain 3-D deformation, combined with the cross-heading tracks synthetic aperture radar (SAR) data. However, because of the limitations of the low earth orbit SAR, a long satellite revisit time, small common areas of the cross-heading tracks data, and the unsatisfied along-track deformation measurement accuracy usually exist in the traditional MAI 3-D deformation retrieval. Geosynchronous SAR (GEO SAR) runs in the geosynchronous orbit, which has the advantages of a large observation area and a short revisit time. This paper focuses on 3-D deformation retrieval by GEO SAR MAI processing. Aiming at the high orbit and the squint looking of GEO SAR, the accurate expressions of the along-track deformation, 3-D deformation, and the errors in GEO SAR MAI processing are given. The distortions and their correction in the MAI interferogram brought by the geometrical difference between the forward- and backward-looking interferograms and the multicycles flat-earth and topographic phases are given. Moreover, an optimal subaperture selection method based on minimum position dilution of precision is proposed. Finally, the effectiveness of the proposed method is validated by simulations and the experiment of BeiDou-2 inclined geosynchronous orbit navigation satellite. The theoretical analysis and the experimental results suggest centimeter-level and even millimeter-level deformation measurement accuracy could be obtained in 3-D by GEO SAR MAI processing.