NeRFPlayer: A Streamable Dynamic Scene Representation with Decomposed Neural Radiance Fields.

Visually exploring in a real-world 4D spatiotemporal space freely in VR has been a long-term quest. The task is especially appealing when only a few or even single RGB cameras are used for capturing the dynamic scene. To this end, we present an efficient framework capable of fast reconstruction, compact modeling, and streamable rendering. First, we propose to decompose the 4D spatiotemporal space according to temporal characteristics. Points in the 4D space are associated with probabilities of belonging to three categories: static, deforming, and new areas. Each area is represented and regularized by a separate neural field. Second, we propose a hybrid representations based feature streaming scheme for efficiently modeling the neural fields. Our approach, coined NeRFPlayer, is evaluated on dynamic scenes captured by single hand-held cameras and multi-camera arrays, achieving comparable or superior rendering performance in terms of quality and speed comparable to recent state-of-the-art methods, achieving reconstruction in 10 seconds per frame and interactive rendering. Project website: https://bit.ly/nerfplayer.

Federated Learning With Privacy-Preserving Ensemble Attention Distillation.

Federated Learning (FL) is a machine learning paradigm where many local nodes collaboratively train a central model while keeping the training data decentralized. This is particularly relevant for clinical applications since patient data are usually not allowed to be transferred out of medical facilities, leading to the need for FL. Existing FL methods typically share model parameters or employ co-distillation to address the issue of unbalanced data distribution. However, they also require numerous rounds of synchronized communication and, more importantly, suffer from a privacy leakage risk. We propose a privacy-preserving FL framework leveraging unlabeled public data for one-way offline knowledge distillation in this work. The central model is learned from local knowledge via ensemble attention distillation. Our technique uses decentralized and heterogeneous local data like existing FL approaches, but more importantly, it significantly reduces the risk of privacy leakage. We demonstrate that our method achieves very competitive performance with more robust privacy preservation based on extensive experiments on image classification, segmentation, and reconstruction tasks.

Sequential multiple assignment randomised trial of a brief contact intervention for suicide risk management among discharged psychiatric patients: an implementation study protocol.

INTRODUCTION: The postdischarge suicide risk among psychiatric patients is significantly higher than it is among patients with other diseases and general population. The brief contact interventions (BCIs) are recommended to decrease suicide risk in areas with limited mental health service resources like China. This study aims to develop a postdischarge suicide intervention strategy based on BCIs and evaluate its implementability under the implementation outcome framework. METHODS AND ANALYSIS: This study will invite psychiatric patients and family members, clinical and community mental health service providers as the community team to develop a postdischarge suicide intervention strategy. The study will recruit 312 patients with psychotic symptoms and 312 patients with major depressive disorder discharged from Shenzhen Kangning Hospital (SKH) in a Sequential Multiple Assignment Randomised Trial. Participants will be initially randomised into two intervention groups to receive BCIs monthly and weekly, and they will be rerandomised into three intervention groups to receive BCIs monthly, biweekly and weekly at 3 months after discharge according to the change of their suicide risk. Follow-ups are scheduled at 1, 3, 6 and 12 months after discharge. With the intention-to-treat approach, generalised estimating equation and survival analysis will be applied. This study will also collect qualitative and quantitative information on implementation and service outcomes from the community team. ETHICS/DISSEMINATION: This study has received ethical approval from the Ethics Committee Review Board of SKH. All participants will provide written informed consent prior to enrolment. The findings of the study will be disseminated through peer-reviewed scientific journals, conference presentations. A project report will be submitted to the National Natural Science Foundation of China as the concluding report of this funded project, and to the mental health authorities in the Shenzhen to refine and apply evidence-based and pragmatic interventions into health systems for postdischarge suicide prevention. TRIAL REGISTRATION NUMBER: NCT04907669.

Nonlocal Low-Rank Tensor Completion for Visual Data.

In this paper, we propose a novel nonlocal patch tensor-based visual data completion algorithm and analyze its potential problems. Our algorithm consists of two steps: the first step is initializing the image with triangulation-based linear interpolation and the second step is grouping similar nonlocal patches as a tensor then applying the proposed tensor completion technique. Specifically, with treating a group of patch matrices as a tensor, we impose the low-rank constraint on the tensor through the recently proposed tensor nuclear norm. Moreover, we observe that after the first interpolation step, the image gets blurred and, thus, the similar patches we have found may not exactly match the reference. We name the problem "Patch Mismatch," and then in order to avoid the error caused by it, we further decompose the patch tensor into a low-rank tensor and a sparse tensor, which means the accepted horizontal strips in mismatched patches. Furthermore, our theoretical analysis shows that the error caused by Patch Mismatch can be decomposed into two components, one of which can be bounded by a reasonable assumption named local patch similarity, and the other part is lower than that using matrix completion. Extensive experimental results on real-world datasets verify our method's superiority to the state-of-the-art tensor-based image inpainting methods.

Study on the Similarity of Biomechanical Behavior between Gelatin and Porcine Liver.

As a natural polymer, gelatin is increasingly being used as a substitute for animals or humans for the simulation and testing of surgical procedures. In the current study, the similarity verification was neglected and a 10 wt.% or 20 wt.% gelatin sample was used directly. To compare the mechanical similarities between gelatin and biological tissues, different concentrations of gelatin samples were subjected to tensile, compression, and indentation tests and compared with porcine liver tissue. The loading rate in the three tests fully considered the surgical application conditions; notably, a loading speed up to 12 mm/s was applied in the indentation testing, the tensile test was performed at a speed of 1 mm/s until fracture, and the compression tests were compressed at a rate of 0.16 mm/s and 1 mm/s. A comparison of the results shows that the mechanical behaviors of low-concentration gelatin samples involved in the study are similar to the mechanical behavior of porcine liver tissue. The results of the gelatin material were mathematically expressed by the Mooney-Rivlin model and the Prony series. The results show that the material properties of gelatin can mimic the range of mechanical characteristics of porcine liver, and gelatin can be used as a matrix to further improve the similarity between substitute materials and biological tissues.

Learning from Synthetic Images via Active Pseudo-Labeling.

Synthetic visual data refers to the data automatically rendered by the mature computer graphic algorithms. With the rapid development of these techniques, we can now collect photo-realistic synthetic images with accurate pixel-level annotations without much effort. However, due to the domain gaps between synthetic data and real data, in terms of not only visual appearance but also label distribution, directly applying models trained on synthetic images to real ones can hardly yield satisfactory performance. Since the collection of accurate labels for real images is very laborious and time-consuming, developing algorithms which can learn from synthetic images is of great significance. In this paper, we propose a novel framework, namely Active Pseudo-Labeling (APL), to reduce the domain gaps between synthetic images and real images. In APL framework, we first predict pseudo-labels for the unlabeled real images in the target domain by actively adapting the style of the real images to source domain. Specifically, the style of real images is adjusted via a novel task guided generative model, and then pseudo-labels are predicted for these actively adapted images. Lastly, we fine-tune the source-trained model in the pseudo-labeled target domain, which helps to fit the distribution of the real data. Experiments on both semantic segmentation and object detection tasks with several challenging benchmark data sets demonstrate the priority of our proposed method compared to the existing state-of-the-art approaches.