Target Distribution Research Articles

Facilitating writing performance of EFL learners via virtual reality: Immersion, presence, embodiment.

Immersive virtual reality (IVR) technology is being used to help EFL learners overcome their difficulties with their language skills, especially writing skills. Past studies showed that the instrument that provides immersion will positively affect learners' written performance. In line with that, this study aims to investigate the vocabulary usage and writing performance of learners who learn vocabulary via IVR versus those who learn from conventional classroom-based instruction. A total of 144 Chinese-speaking English learners, who were divided into experimental group (N = 69) and control group (N = 75), experienced the treatments related to the writing tasks. The results show that the learners in the experimental group wrote more informatively and presented more details. Comparative analysis revealed that learners using IVR performed significantly better on target word usage, lexical density, distribution richness, and completion of task than those in the conventional classroom. Based on the results, it would seem that the positive transfer of learning may be related to the experience of exploring in a virtual environment. The immersion of IVR and the sense of presence and embodiment enable learners to benefit from their immersive experience which aids the use of vocabulary in their writing. The implication of the study demonstrated the impact of the technological factors, whereby what causes the improvement in writing performance is due to the learners' virtual experience and their sense of embodiment.

Finite mixture modeling in time series: A survey of Bayesian filters and fusion approaches

From the celebrated Gaussian mixture, model averaging estimators to the cutting-edge multi-Bernoulli mixture of various forms, finite mixture models offer a fundamental and flexible means to deal with uncertainties arisen in the estimation and learning realm. In the context of recursive estimation, both the uncertainties due to model maneuvering and multi-target data association derive a need for representing the single/multiple-target probability distribution by a finite mixture described by a number of parameters that are iteratively updated over time in the framework of Bayesian inference, which we call generally the Bayesian mixture (BM) filtering. In addition, density fusion between netted agents/sensors may be addressed by linearly combining their local estimations which are often correlated with each other in a complicated means, leading to a fused mixture too. In either case, the final estimate is given by the arithmetic average (AA) of all the weighted components in the mixture, regarding either a single target or multiple targets. Along with this are versatile mixture adaption and optimization strategies which aim to use a small number of components for the best fit to the target distribution. This survey provides a comprehensive overview of representative single/multiple-target BM filters and fusion approaches with the use of either a single sensor or multiple sensors in a unified and coherent fashion. State-of-the-art, relevant adaption and optimization technologies and remaining challenges are discussed. All strive to gain more insights of the approach.

A Lightweight Object Detection Algorithm for Remote Sensing Images Based on Attention Mechanism and YOLOv5s

The specific characteristics of remote sensing images, such as large directional variations, large target sizes, and dense target distributions, make target detection a challenging task. To improve the detection performance of models while ensuring real-time detection, this paper proposes a lightweight object detection algorithm based on an attention mechanism and YOLOv5s. Firstly, a depthwise-decoupled head (DD-head) module and spatial pyramid pooling cross-stage partial GSConv (SPPCSPG) module were constructed to replace the coupled head and the spatial pyramid pooling-fast (SPPF) module of YOLOv5s. A shuffle attention (SA) mechanism was introduced in the head structure to enhance spatial attention and reconstruct channel attention. A content-aware reassembly of features (CARAFE) module was introduced in the up-sampling operation to reassemble feature points with similar semantic information. In the neck structure, a GSConv module was introduced to maintain detection accuracy while reducing the number of parameters. Experimental results on remote sensing datasets, RSOD and DIOR, showed an improvement of 1.4% and 1.2% in mean average precision accuracy compared with the original YOLOv5s algorithm. Moreover, the algorithm was also tested on conventional object detection datasets, PASCAL VOC and MS COCO, which showed an improvement of 1.4% and 3.1% in mean average precision accuracy. Therefore, the experiments showed that the constructed algorithm not only outperformed the original network on remote sensing images but also performed better than the original network on conventional object detection images.

PO-03-090 NOVEL HIGH RESOLUTION MR IMAGING IMPROVES PRE-PROCEDURE IDENTIFICATION OF ABLATION TARGETS IN POST-INFARCT VENTRICULAR TACHYCARDIA

MRI-based personalized virtual heart models (VHM) are a useful tool for reconstructing scar distribution and identifying VT circuits for planning catheter ablation procedure, an effective therapy for preventing sudden cardiac death from VT in patients with myocardial infarction. However, the MRI resolution used to reconstruct VHMs may not be sufficient to accurately detect all VT circuits, possibly leading to erroneous predictions. To predict catheter ablation targets in the arrhythmogenic substrate using VHMs reconstructed from both clinical and high-resolution 2D MRI datasets and compare the predictions. High-resolution (1.352 x 3mm) short-axis cardiac MRI were acquired by combining two subsequent clinical resolution (1.352 x 6mm) short-axis MRI from nine post-infarct patients undergoing VT ablation. The high-resolution and two clinical MRI were used to reconstruct a total of three personalized VHMs for each patient that represented the patient-specific infarct remodeling. Rapid pacing was used to assess the VT circuits and identify ablation targets in each VHM. Differences in VT inducibility and location were assessed between the high and clinical resolution VHM reconstructions, as well as between the two clinical resolution VHM reconstructions. Virtual ablations were performed to identify emergent VTs and were also compared between the VHMs for each patient. The amount of fibrotic remodeling changed by 12.6 ± 3.3% between the high-resolution and two clinical MRI VHMs for each patient, while the distribution was only moderately affected (Fig 1). The heterogeneity of the myocardium and the fractal dimension (complexity) of the scar tissue was not statistically different between the VHMs for each patient (Fig 1). VT inducibility was observed in all VHMs with an average of 2.7 [1-4] unique VT morphologies induced in each VHM. In two patients, only one clinical VHM was able to identify all the unique VT morphologies observed in the high-resolution VHM. VT exit sites in the clinical MRI VHMs were within 5mm of the exit sites observed in the high-resolution VHMs in 72.2% of cases. Virtual ablations uncovered emergent VT circuits in two of the high-resolution VHMs, compared to only one of the corresponding clinical MRI VHMs. The distribution of ablation targets required to terminate VT was statistically different in 44.4% of cases (Fig 1). VHMs reconstructed from high-resolution MRI more accurately identify VT circuits and may improve catheter ablation outcomes.

OC-0783 treatment outcome prediction from modeling the clinical target distribution for high-grade gliomas

OC-0783 treatment outcome prediction from modeling the clinical target distribution for high-grade gliomas

Frequency-based pseudo-domain generation for domain generalizable object detection

Domain generalizable object detection (DGOD) aims to train a detector that performs well on multiple unseen target domains, which is crucial for deploying the detector in practice. Recent methods for DGOD typically inherit the idea from domain adaptation to align or disentangle features, but these methods struggle to handle unknown target distributions. In this paper, we propose a unified framework to tackle the DGOD task from a novel pseudo-domain generation perspective. Our framework comprises two stages: distribution diversification and domain-invariant feature learning. In the distribution diversification stage, we design a Frequency-based Pseudo-domain Generator (FPG) to construct the pseudo domain via excavating latent style information and enhancing semantic information in frequency space. The generated pseudo domain can provide diverse training distributions, which enhances generalization performance. In the domain-invariant feature learning stage, we introduce Rotation Prediction and Semantic Consistency (RPSC) learning, including an auxiliary self-supervised task rotation prediction to encourage generalized feature learning and a semantic consistency loss to enforce the detector to be invariant of domain shifts. Extensive experiments are conducted on various object detection benchmarks, demonstrating the superiority of our approach over state-of-the-art methods in both single-source and multi-source settings.

Wasserstein Adversarially Regularized Graph Autoencoder

This paper introduces Wasserstein Adversarially Regularized Graph Autoencoder (WARGA), an implicit generative algorithm that directly regularizes the latent distribution of node embedding to a target distribution via the Wasserstein metric. To ensure the Lipschitz continuity, we propose two approaches: WARGA-WC that uses weight clipping method and WARGA-GP that uses gradient penalty method. The proposed models have been validated by link prediction and node clustering on real-world graphs with visualizations of node embeddings, in which WARGA generally outperforms other state-of-the-art models based on Kullback–Leibler (KL) divergence and typical adversarial framework.

Scalable Bayesian Transport Maps for High-Dimensional Non-Gaussian Spatial Fields

A multivariate distribution can be described by a triangular transport map from the target distribution to a simple reference distribution. We propose Bayesian nonparametric inference on the transport map by modeling its components using Gaussian processes. This enables regularization and uncertainty quantification of the map estimation, while resulting in a closed-form and invertible posterior map. We then focus on inferring the distribution of a nonstationary spatial field from a small number of replicates. We develop specific transport-map priors that are highly flexible and are motivated by the behavior of a large class of stochastic processes. Our approach is scalable to high-dimensional distributions due to data-dependent sparsity and parallel computations. We also discuss extensions, including Dirichlet process mixtures for flexible marginals. We present numerical results to demonstrate the accuracy, scalability, and usefulness of our methods, including statistical emulation of non-Gaussian climate-model output. Supplementary materials for this article are available online.

Manifold lifting: scaling Markov chain Monte Carlo to the vanishing noise regime

Abstract Standard Markov chain Monte Carlo methods struggle to explore distributions that concentrate in the neighbourhood of low-dimensional submanifolds. This pathology naturally occurs in Bayesian inference settings when there is a high signal-to-noise ratio in the observational data but the model is inherently over-parametrised or nonidentifiable. In this paper, we propose a strategy that transforms the original sampling problem into the task of exploring a distribution supported on a manifold embedded in a higher-dimensional space; in contrast to the original posterior this lifted distribution remains diffuse in the limit of vanishing observation noise. We employ a constrained Hamiltonian Monte Carlo method, which exploits the geometry of this lifted distribution, to perform efficient approximate inference. We demonstrate in numerical experiments that, contrarily to competing approaches, the sampling efficiency of our proposed methodology does not degenerate as the target distribution to be explored concentrates near low-dimensional submanifolds. Python code reproducing the results is available at https://doi.org/10.5281/zenodo.6551654.

Statistical testing under distributional shifts

AbstractWe introduce statistical testing under distributional shifts. We are interested in the hypothesis P*∈H0 for a target distribution P*, but observe data from a different distribution Q*. We assume that P* is related to Q* through a known shift τ and formally introduce hypothesis testing in this setting. We propose a general testing procedure that first resamples from the observed data to construct an auxiliary data set (similarly to sampling importance resampling) and then applies an existing test in the target domain. We prove that if the size of the resample is of order o(n) and the resampling weights are well behaved, this procedure inherits the pointwise asymptotic level and power from the target test. If the map τ is estimated from data, we maintain the above guarantees under mild conditions on the estimation. Our results extend to finite sample level, uniform asymptotic level, a different resampling scheme, and statistical inference different from testing. Testing under distributional shifts allows us to tackle a diverse set of problems. We argue that it may prove useful in contextual bandit problems and covariate shift, show how it reduces conditional to unconditional independence testing and provide example applications in causal inference.

Stein variational gradient descent with learned direction

Recently, as a particle-based Bayesian inference method, Stein variational gradient descent (SVGD) has attracted much attention due to its strong approximation power and efficiency compared to traditional variational inference and Markov chain Monte Carlo sampling methods. However, the original SVGD method assumes the descent direction to be within an reproducing kernel Hilbert space (RKHS), which on one hand limits its expressive power, and on the other hand suffers from the curse of dimensionality. To address this issue, we propose to parameterize the descent direction using a flexible neural network. With this parameterization strategy, the descent direction can be possibly optimized in each step for updating the particles, such that the target distribution can be expected to be better approximated. Experiments on both synthetic and real datasets demonstrate the effectiveness and efficiency of the proposed method, especially in the high-dimensional situations, as compared with the original SVGD.

Implications of using the clinical target distribution as voxel-weights in radiation therapy optimization

Objective. Delineating and planning with respect to regions suspected to contain microscopic tumor cells is an inherently uncertain task in radiotherapy. The recently proposed clinical target distribution (CTD) is an alternative to the conventional clinical target volume (CTV), with initial promise. Previously, using the CTD in planning has primarily been evaluated in comparison to a conventionally defined CTV. We propose to compare the CTD approach against CTV margins of various sizes, dependent on the threshold at which the tumor infiltration probability is considered relevant. Approach. First, a theoretical framework is presented, concerned with optimizing the trade-off between the probability of sufficient target coverage and the penalties associated with high dose. From this framework we derive conventional CTV-based planning and contrast it with the CTD approach. The approaches are contextualized further by comparison with established methods for managing geometric uncertainties. Second, for both one- and three-dimensional phantoms, we compare a set of CTD plans created by varying the target objective function weight against a set of plans created by varying both the target weight and the CTV margin size. Main results. The results show that CTD-based planning gives slightly inefficient trade-offs between the evaluation criteria for a case in which near-minimum target dose is the highest priority. However, in a case when sparing a proximal organ at risk is critical, the CTD is better at maintaining sufficiently high dose toward the center of the target. Significance. We conclude that CTD-based planning is a computationally efficient method for planning with respect to delineation uncertainties, but that the inevitable effects on the dose distribution should not be disregarded.

A novel automated planning approach for multi-anatomical sites cancer in Raystation treatment planning system

PurposeTo develop an automated planning approach in Raystation and evaluate its feasibility in multiple clinical application scenarios. MethodsAn automated planning approach (Ruiplan) was developed by using the scripting platform of Raystation. Radiotherapy plans were re-generated both automatically by using Ruiplan and manually. 60 patients, including 20 patients with nasopharyngeal carcinoma (NPC), 20 patients with esophageal carcinoma (ESCA), and 20 patients with rectal cancer (RECA) were retrospectively enrolled in this study. Dosimetric and planning efficiency parameters of the automated plans (APs) and manual plans (MPs) were statistically compared. ResultsFor target coverage, APs yielded superior dose homogeneity in NPC and RECA, while maintaining similar dose conformity for all studied anatomical sites. For OARs sparing, APs led to significant improvement in most OARs sparing. The average planning time required for APs was reduced by more than 43% compared with MPs. Despite the increased monitor units (MUs) for NPC and RECA in APs, the beam-on time of APs and MPs had no statistical difference. Both the MUs and beam-on time of APs were significantly lower than that of MPs in ESCA. ConclusionsThis study developed a new automated planning approach, Ruiplan, it is feasible for multi-treatment techniques and multi-anatomical sites cancer treatment planning. The dose distributions of targets and OARs in the APs were similar or better than those in the MPs, and the planning time of APs showed a sharp reduction compared with the MPs. Thus, Ruiplan provides a promising approach for realizing automated treatment planning in the future.

An unsupervised domain adaptation deep learning method for spatial and temporal transferable crop type mapping using Sentinel-2 imagery

Accurate crop type mapping is essential for crop growth monitoring and yield estimation. Recently, various machine learning methods have been increasingly used for crop type mapping, but they often lose their validity when directly applied to other regions and years due to differences in the distribution of source and target data, that is, domain shift. To address the problem, we developed a deep adaptation crop classification network (DACCN) based on the idea of unsupervised domain adaptation (UDA). The proposed DACCN mainly consists of two parts, a feature extractor that converts the original input into high-level representations, and a domain aligner where the discrepancy between source and target distributions is measured using multiple kernel variant of maximum mean discrepancy (MK-MMD). Four states in the United States (U.S.) Corn Belt and two provinces in northeastern China were used as study areas, where samples used for model building and accuracy evaluation were collected based on time-series Sentinel-2 imagery and reference maps in 2018 and 2019. Then, three experiments were designed to verify the transferability of DACCN across space, time, and space–time, respectively. In each experiment, the proposed DACCN was compared to deep crop classification network (DCCN), a model with a similar structure to DACCN but without the domain adaptation mechanism, and two machine learning methods, random forest (RF) and support vector machines (SVM). The experimental results showed that DACCN outperformed other models in most transfer cases with overall classification accuracies ranging from 0.835 to 0.922. The DACCN also performed better in spatially continuous mapping with its predicted crop type maps more consistent with the reference ones. As an innovative application of transfer learning in crop type mapping, the methodology proposed in this study effectively addressed the problem of missing labels in target domains and alleviated the negative impact of domain shift.

Exploring personalization via federated representation Learning on non-IID data

Federated Learning (FL) can learn a global model across decentralized data over different clients. However, it is susceptible to statistical heterogeneity of client-specific data. Clients focus on optimizing for their individual target distributions, which would yield divergence of the global model due to inconsistent data distributions. Moreover, federated learning approaches adhere to the scheme of collaboratively learning representations and classifiers, further exacerbating such inconsistency and resulting in imbalanced features and biased classifiers. Hence, in this paper, we propose an independent two-stage personalized FL framework, i.e., Fed-RepPer, to separate representation learning from classification in federated learning. First, the client-side feature representation models are learned using supervised contrastive loss, which enables local objectives consistently, i.e., learning robust representations on distinct data distributions. Local representation models are aggregated into the common global representation model. Then, in the second stage, personalization is studied by learning different classifiers for each client based on the global representation model. The proposed two-stage learning scheme is examined in lightweight edge computing that involves devices with constrained computation resources. Experiments on various datasets (CIFAR-10/100, CINIC-10) and heterogeneous data setups show that Fed-RepPer outperforms alternatives by utilizing flexibility and personalization on non-IID data.

A Multi-Frame Superposition Detection Method for Dim-Weak Point Targets Based on Optimized Clustering Algorithm

Long-range surveillance and early warning of space targets are significant factors in space security. Under remote observation conditions, the energy performance of the target is weak and subject to environmental and imaging process contamination. Most detection methods are aimed at targets with a high signal-to-noise ratio (SNR) or local contrast, and the detection performance for dim-weak small targets is poor; therefore, the target signal is often enhanced by energy accumulation. However, owing to the interference caused by the instability of the imaging system, energy accumulation errors occur in the target, resulting in the dispersion of the target energy, making detection a challenge. To solve the above problem, this study proposed a multi-frame superposition detection method for dim-weak point targets based on an optimized clustering algorithm by combining the clustering method with the inherent features of the target and using the difference between the target and noise energy distribution for detection. First, we simulated the multi-frame imaging process of the target post-disturbance and established an optical imaging system model of the dim-weak target. Subsequently, we used data dimension reduction and outlier removal to extract the target potential area. Finally, the data were sent to the clustering model for calculation and judgment. Given that the accuracy rate reaches 87.1% when the SNR is 1 dB, the experimental results show that the detection method proposed in this paper can effectively detect dim-weak targets with low SNR. In addition, there is a significant improvement in the detection performance of the receiver characteristic curve compared with other algorithms in the real scene, which further proves the superiority of the method in this paper.

Is It Who You Are or Where You Are? Accounting for Compositional Differences in Cross-Site Treatment Effect Variation

In multisite trials, learning about treatment effect variation across sites is critical for understanding where and for whom a program works. Unadjusted comparisons, however, capture “compositional” differences in the distributions of unit-level features as well as “contextual” differences in site-level features, including possible differences in program implementation. Our goal in this article is to adjust site-level estimates for differences in the distribution of observed unit-level features: If we can reweight (or “transport”) each site to have a common distribution of observed unit-level covariates, the remaining treatment effect variation captures contextual and unobserved compositional differences across sites. This allows us to make apples-to-apples comparisons across sites, parceling out the amount of cross-site effect variation explained by systematic differences in populations served. In this article, we develop a framework for transporting effects using approximate balancing weights, where the weights are chosen to directly optimize unit-level covariate balance between each site and the common target distribution. We first develop our approach for the general setting of transporting the effect of a single-site trial. We then extend our method to multisite trials, assess its performance via simulation, and use it to analyze a series of multisite trials of adult education and vocational training programs. In our application, we find that distributional differences are potentially masking cross-site variation. Our method is available in the balancer R package.

Tumor-Agnostic Precision Medicine from the AACR GENIE Database: Clinical implications.

Biomarker-driven cancer therapy has revolutionized precision oncology. With a better understanding of tumor biology, tissue-agnostic targets have been characterized and explored which ultimately led to therapeutics with pan-cancer efficacy. To date, five molecular biomarkers have obtained FDA tissue-agnostic approval for targeted therapies and immunotherapies. Those include BRAF V600E mutations, RET fusions, NTRK fusions, high tumor mutation burden (TMB), and deficient mismatch repair (dMMR/MSI-High). Herein, we have used data from AACR project GENIE to explore the clinico-genomic landscape of these alterations. AACR GENIE is a publicly accessible registry of genomic data from multiple collaborating cancer centers. Current database (version 13.0) includes sequencing data of 168,423 samples collected from patients with different cancers. We were able to identify BRAF V600E, RET fusions, NTRK fusions, and high TMB in 2.9%, 1.6%, 1.5%, and 15.2% of pan-cancer samples; respectively. In this article, we describe the distribution of those tissue-agnostic targets among different cancer types. In addition, we summarize the current prospect on the biology of these alterations and evidence on approved drugs including pembrolizumab, dostarilmab, larotrectinib, entrectinib, selpercatinib, and dabrafenib/trametinib combination.

Sampling by divergence minimization

We introduce a Markov Chain Monte Carlo (MCMC) method that is designed to sample from target distributions with irregular geometry using an adaptive scheme. In cases where targets exhibit non-Gaussian behavior, we propose that adaptation should be regional rather than global. Our algorithm minimizes the information projection component of the Kullback–Leibler (KL) divergence between the proposal and target distributions to encourage proposals that are distributed similarly to the regional geometry of the target. Unlike traditional adaptive MCMC, this procedure rapidly adapts to the geometry of the target’s current position as it explores the surrounding space without the need for many preexisting samples. The divergence minimization algorithms are tested on target distributions with irregularly shaped modes and we provide results demonstrating the effectiveness of our methods.

Dynamic-Pix2Pix: Medical image segmentation by injecting noise to cGAN for modeling input and target domain joint distributions with limited training data

Learning to translate images from a source to a target domain with applications such as converting simple line drawing to oil painting has attracted significant attention. The quality of translated images is directly related to two crucial issues. First, the consistency of the output distribution with that of the target is essential. Second, the generated output should have a high correlation with the input. Conditional Generative Adversarial Networks, cGANs, are the most common models for translating images. The performance of a cGAN drops when we use a limited training dataset. In this work, we study the obstacles that prevent a U-shaped model from learning the target domain distribution from limited data by using noise as input. This study helps to increase the Pix2Pix (a form of cGAN) target distribution modeling ability from limited data with the help of dynamic neural network theory. Our model has two learning cycles. The model learns the correlation between input and ground truth in the first cycle. Then, the model's architecture is refined in the second cycle to learn the target distribution from noise input. These processes are executed in each iteration of the training procedure. Helping the cGAN learn the target distribution from noise input results in a better model generalization during the test time and allows the model to fit almost perfectly to the target domain distribution. As a result, our model surpasses the Pix2Pix model in segmenting HC18 and Montgomery's chest x-ray images. Both qualitative and Dice scores show the superiority of our model. Although our proposed method does not use thousand of additional data for pretraining, it produces comparable results for the in and out-domain generalization compared to the state-of-the-art methods.