LoCo: Low-Bit Communication Adaptor for Large-Scale Model Training.

To efficiently train large-scale models, low-bit gradient communication compresses full-precision gradients on local GPU nodes into low-precision ones for higher gradient synchronization efficiency among GPU nodes. However, it often degrades training quality due to compression information loss. To address this, we propose the Low-bit Communication Adaptor (LoCo), which compensates gradients on local GPU nodes before compression, ensuring efficient synchronization without compromising training quality.

Real-Time Autoregressive Forecast of Cardiac Features for Psychophysiological Applications.

Forecasting the near-exact moments of cardiac phases is crucial for several cardiovascular health applications. For instance, forecasts can enable the timing of specific stimuli (e.g.

Resolving Symmetry Ambiguity in Correspondence-based Methods for Instance-level Object Pose Estimation.

Estimating the 6D pose of an object from a single RGB image is a critical task that becomes additionally challenging when dealing with symmetric objects. Recent approaches typically establish one-to-one correspondences between image pixels and 3D object surface vertices. However, the utilization of one-to-one correspondences introduces ambiguity for symmetric objects.

GRATCR: Epitope-Specific T Cell Receptor Sequence Generation With Data-Efficient Pre-Trained Models.

T cell receptors (TCRs) play a crucial role in numerous immunotherapies targeting tumor cells. However, their acquisition and optimization present significant challenges, involving laborious and time-consuming wet lab experimental resource. Deep generative models have demonstrated remarkable capabilities in functional protein sequence generation, offering a promising solution for enhancing the acquisition of specific TCR sequences.

HATNet: EEG-Based Hybrid Attention Transfer Learning Network for Train Driver State Detection.

Electroencephalography (EEG) is widely utilized for train driver state detection due to its high accuracy and low latency. However, existing methods for driver status detection rarely use the rich physiological information in EEG to improve detection performance. Moreover, there is currently a lack of EEG datasets for abnormal states of train drivers.

Gait Recognition in the Wild: A Large-scale Benchmark and NAS-based Baseline.

Gait benchmarks empower the research community to train and evaluate high-performance gait recognition systems. Even though growing efforts have been devoted to cross-view recognition, academia is restricted by current existing databases captured in the controlled environment. In this paper, we contribute a new benchmark and strong baseline for Gait REcognition in the Wild (GREW).

Biologically Interpretable Model for Precise Recurrence Prediction of Non-Small Cell Lung Cancer.

The precise prediction of preoperative recurrence in non-small cell lung cancer (NSCLC) that is suitable for clinical application is still an open question. Recent advancements integrating genomic data with deep learning have shown promise in enhancing recurrence analysis in NSCLC patients. However, the lack of interpretability in the decision-making process of DNN models has hindered their clinical trustworthiness.

Fast Diffeomorphic Image Registration using Patch based Fully Convolutional Networks.

Diffeomorphic image registration is a fundamental step in medical image analysis, owing to its capability to ensure the invertibility of transformations and preservation of topology. Currently, unsupervised learning-based registration techniques primarily extract features at the image level, potentially limiting their efficacy. This paper proposes a novel unsupervised learning-based fully convolutional network (FCN) framework for fast diffeomorphic image registration, emphasizing feature acquisition at the image patch level.

STFormer: Learning to Explore Spot Relationships for Spatial Transcriptomics Prediction from Histology of Colorectal Cancer.

Spatial transcriptomics (ST) offers insights into gene expression patterns within tumor microenvironments, but its widespread application is impeded by cost constraints. To address this, predicting ST from Histology emerges as a cost-effective alternative. However, current methods such as STNet, HistoGene, and Hist2ST exhibit limitations, either overlooking stain variation across datasets or failing to well explore inter-spot correlations in scenarios with limited Whole Slide Image (WSI) data.

A simplified method for long-term maintenance of human induced pluripotent stem-cell derived neural networks.

In-vitro models of neuronal networks have become a powerful tool for modeling network activity in the human brain. The exploration of network properties has largely been made possible via microelectrode arrays (MEAs). However, addressing certain tissue engineering challenges remains imperative for their long-term utilization.

Revealing Fine-grained Genetically Informed Cortical Parcellation Maps of Neonates Based on Multi-view Spectral Clustering.

Genetic factors have been proven to be one of the major determinants in shaping the neonatal cerebral cortex. Previous research has demonstrated distinct genetic influences on the spatial patterns of cortical thickness (CT) and surface area (SA) in neonates, leading to their unique genetically informed parcellation maps. However, these parcellation maps were derived at coarse scales and only reliant on single cortical properties, making them unable to comprehensively characterize the fine-grained genetically regulated patterns of the neonatal cerebral cortex.

Elbow Joint Classification for Total Elbow Arthroplasty.

Current elbow prosthesis design and fabrication is not catering for the target population, leading to a prevalent occurrence of prosthesis mismatch in total elbow arthroplasty (TEA) surgeries. To address this challenge, it is crucial to develop an efficient elbow joint classification method that accurately captures specific morphological variations, advancing the design of diverse prostheses for optimal matching. In this paper, we introduce two classification algorithms utilizing shape features extracted from the Riemannian manifold and anatomical features derived from three-dimensional (3D) measurements.

GANs-guided Conditional Diffusion Model for Synthesizing Contrast-enhanced Computed Tomography Images.

In contrast to non-contrast computed tomography (NC-CT) scans, contrast-enhanced (CE) CT scans can highlight discrepancies between abnormal and normal areas, commonly used in clinical diagnosis of focal liver lesions. However, the use of contrast agents in CE-CT scans imposes significant physical and economic burdens on patients in clinical practice. Recently, Generative Adversarial Networks (GANs)-based synthesis models offer an alternative approach that obtains CE-CT images from NC-CT images.

Hierarchical Auto-labeling of Coronary Arteries on CT Coronary Angiography Images.

The auto-labeling of coronary artery segments plays an important role in the diagnosis of cardiovascular diseases. Due to the high degree of complexity and diversity in coronary artery structures, it is still a very challenging task after many years of exploration and study. In this paper, we propose a hierarchical scheme based on PointNet++ models and new topological structural features for automatic labeling of coronary artery segments.

A Three-Stage Semi-Supervised Learning Approach to Spine Image Segmentation.

Spine segmentation in computed tomography (CT) images is critical for automatic analysis, especially when focusing on varied spinal anatomy. Despite having comprehensive annotations for normal vertebrae, many datasets do not encompass labeled fracture data, posing challenges for predictive modeling. This research presents a three-stage 2.

UPMatch: Enhancing Semi-Supervised Medical Image Classification through Contrastive Learning with Unreliable Pseudo Labels.

Pseudo-labeling based semi-supervised learning (SSL) framework has proven highly successful in medical image analysis (MIA) by addressing the problem of a shortage of labeled samples. However, the existing SSL methods use a fixed or flexible confidence threshold to filter reliable samples, leaving large number of unlabeled samples unused. This is a more serious issue in MIA because of the low inter-class distance and imbalanced categories.

Effects of 3D Stimuli With Frequency Ranges, Patterns, and Shapes on SSVEP-BCI Performance in Virtual Reality.

Traditional steady-state visual evoked potential (SSVEP) based brain-computer interface (BCI) systems offer stability and simplicity in evoking brain responses, but their practical utility is limited by immovable screens for visual stimuli. Virtual Reality (VR) technology provides a more natural and immersive environment to evoke SSVEP signals. However, the design methods for visual stimuli in VR environments remain to be explored, especially under the stereoscopic vision conditions.

Line-of-Sight Depth Attention for Panoptic Parsing of Distant Small-Faint Instances.

Current scene parsers have effectively distilled abstract relationships among refined instances, while overlooking the discrepancies arising from variations in scene depth. Hence, their potential to imitate the intrinsic 3D perception ability of humans is constrained. In accordance with the principle of perspective, we advocate first grading the depth of the scenes into several slices, and then digging semantic correlations within a slice or between multiple slices.

The dual missions of FoxO3a in inflammatory diseases: Regulation of antioxidant enzymes and involvement in programmed cell death.

The transcription factor FoxO3a plays a crucial role in the process of cells adapting to various stress conditions. Multiple post - translational modifications and epigenetic mechanisms work together to precisely regulate the activity of FoxO3a, influencing its subcellular localization, stability, interactions with other proteins, DNA - binding affinity, and transcriptional regulatory capacity. Under different chemical signal stimuli and subcellular environments, the activation of FoxO3a triggered by oxidative stress can initiate diverse transcriptional programs, which are essential for the body to resist oxidative damage.

BTEX exposure and metabolite levels in Taiwan schoolchildren near petrochemical areas.

Introduction: Benzene, toluene, ethylbenzene, and xylenes (BTEX) in ambient air pose significant health risks for residents near petrochemical facilities. However, limited research has investigated the correlation between BTEX exposure and urinary metabolites in children. This is the first study to determine this association among primary school children near petrochemical industrial parks (PIPs) in Taiwan.

Fall-Risk Monitoring in Diverse Terrains Using Dual-Task Learning and Wearable Sensing System.

As the elderly population grows, falling accidents become more frequent, and the need for fall-risk monitoring systems increases. Deep learning models for fallrisk movement detection neglect the connections between the terrain and fall-hazard movements. This issue can result in false alarms, particularly when a person encounters changing terrain.

Coarse for Fine: Bounding Box Supervised Thyroid Ultrasound Image Segmentation Using Spatial Arrangement and Hierarchical Prediction Consistency.

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.

Autonomous clustering by fast find of mass and distance peaks.

Clustering is an essential analytical tool across a wide range of scientific fields, including biology, chemistry, astronomy, and pattern recognition. This paper introduces a novel clustering algorithm as a competitive alternative to existing methods, based on the intuitive principle that a cluster should merge with its nearest neighbor with a higher mass, unless both clusters have relatively large masses and the distance between them is also substantial. By identifying peaks in mass and distance, the algorithm effectively detects and removes incorrect mergers.

Sparsity-Aware Distributed Learning for Gaussian Processes With Linear Multiple Kernel.

Gaussian processes (GPs) stand as crucial tools in machine learning and signal processing, with their effectiveness hinging on kernel design and hyperparameter optimization. This article presents a novel GP linear multiple kernel (LMK) and a generic sparsity-aware distributed learning framework to optimize the hyperparameters. The newly proposed grid spectral mixture product (GSMP) kernel is tailored for multidimensional data, effectively reducing the number of hyperparameters while maintaining good approximation capability.

Dual-Camera All-in-Focus Neural Radiance Fields.

We present the first framework capable of synthesizing the all-in-focus neural radiance field (NeRF) from inputs without manual refocusing. Without refocusing, the camera will automatically focus on the fixed object for all views, and current NeRF methods typically using one camera fail due to the consistent defocus blur and a lack of sharp reference. To restore the all-in-focus NeRF, we introduce the dual-camera from smartphones, where the ultra-wide camera has a wider depth-of-field (DoF) and the main camera possesses a higher resolution.