Statistical Alignment Research Articles

Cross-Domain Adaptation Techniques for Robust Plant Disease Detection: A DANN-CORAL Hybrid Approach

Plant disease detection with deep learning models has shown promising results, but these models often struggle with generalizing across diverse agricultural environments due to domain shifts in imaging conditions. This paper presents a novel hybrid approach focusing on cross-domain adaptation techniques to address the challenge of domain shift. Our proposed method combines the Domain-Adversarial Neural Network (DANN) with Correlation Alignment (CORAL) to mitigate domain shifts between datasets. The DANN framework enforces domain-invariant feature learning through adversarial training. Using the PlantVillage Dataset, with controlled environment images, and the New Plant Village Dataset, with varied conditions, the model is first trained on PlantVillage and then adapted to New Plant Village using the CORAL loss to support the second-order statistics. In case of domain shift experiments with various datasets, DANN-CORAL achieved accuracies 91.39%, precision 93.36%, recall 88.9% and F1-scores 91.05% indicating the robustness and generalizability of our model is better than the other baseline models. This approach enhances model robustness and adaptability, providing insights into combining adversarial and statistical alignment for cross-domain adaptation in agricultural imaging.

Implementation of the ARIMA model for prediction of economic variables: evidence from the health sector in Brazil

In recent decades, quantitative models for forecasting economic crises have garnered significant interest from researchers, policymakers, and public and private institutions. Identifying the most appropriate models for predicting economic time series behaviors during crises is a pressing challenge. Effective techniques can be instrumental in forestalling financial irregularities, thus empowering institutions to deploy remedial actions and swiftly minimizing further economic setbacks. Contemporary literature introduces various forecasting models, such as the autoregressive integrated moving average (ARIMA) model. Recognized for its statistical alignment with numerous linear models, the ARIMA model has demonstrated its efficacy across various domains. This paper delves into applying the ARIMA model to predict five critical economic time series that substantially influenced Brazil’s public and private healthcare sectors throughout the economic crisis between 2000 and 2020. These time series encompassed the variables (i) the gross domestic product—GDP, (ii) the Extended National Consumer Price Index—IPCA, (iii) the unemployment rate, (iv) the total number of health plan beneficiaries, and (v) total number of individual health plan beneficiaries. Importantly, this study provides a comprehensive outline of the ARIMA implementation process, underscoring that precise forecasting is pivotal for managers aiming to curtail financial anomalies and avert resource shortages. The findings highlight the ARIMA model’s (1, 0, 2), (2, 2, 1), (0, 1, 2), (1, 1, 2), and (2, 2, 1) viability in accurately forecasting health-related time series, exceeding 95% accuracy for economic variables analyzed. These results have significant practical implications for healthcare managers and decision-makers. By offering accurate forecasts of critical economic metrics, such as the unemployment rate and the transition of beneficiaries between public and private health systems during economic downturns, this research provides valuable insights for strategic planning within the healthcare sector.

COATi: Statistical Pairwise Alignment of Protein-Coding Sequences.

Sequence alignment is an essential method in bioinformatics and the basis of many analyses, including phylogenetic inference, ancestral sequence reconstruction, and gene annotation. Sequencing artifacts and errors made during genome assembly, such as abiological frameshifts and incorrect early stop codons, can impact downstream analyses leading to erroneous conclusions in comparative and functional genomic studies. More significantly, while indels can occur both within and between codons in natural sequences, most amino-acid- and codon-based aligners assume that indels only occur between codons. This mismatch between biology and alignment algorithms produces suboptimal alignments and errors in downstream analyses. To address these issues, we present COATi, a statistical, codon-aware pairwise aligner that supports complex insertion-deletion models and can handle artifacts present in genomic data. COATi allows users to reduce the amount of discarded data while generating more accurate sequence alignments. COATi can infer indels both within and between codons, leading to improved sequence alignments. We applied COATi to a dataset containing orthologous protein-coding sequences from humans and gorillas and conclude that 41% of indels occurred between codons, agreeing with previous work in other species. We also applied COATi to semiempirical benchmark alignments and find that it outperforms several popular alignment programs on several measures of alignment quality and accuracy.

A novel deep transfer learning framework integrating general and domain-specific features for EEG-based brain–computer interface

The statistical variability of electroencephalographic (EEG) signals across individuals poses a common challenge for brain–computer interfaces (BCI). Specifically, the reuse of pre-recorded data from previous subjects presents a significant obstacle in decoding new subjects. To address these challenges, this paper introduces EEGTransferNet, an end-to-end modular transfer learning framework that enhances the similarity of statistical distributions across subjects. It utilizes convolutional neural networks as backbones for feature extraction and incorporates statistical distribution alignment and domain adaptation in different network layers to learn general features and domain-specific features respectively. The robust generalization design of EEGTransferNet enables it to handle multiple BCI paradigms using a single network making it efficient and effective. Its modular structure facilitates customization of backbone and transfer modules for diverse tasks. The optimal structure of EEGTransferNet was determined through analysis and extensive experimentation. We validated it on motor imagery (MI), error-related negativity (ERN), and rapid serial visual presentation (RSVP) datasets, which exhibits competitive performance compared to state-of-the-art methods. The experimental results confirm the effectiveness and superiority of EEGTransferNet.

MNEMONIC: Multikernel contrastive domain adaptation for time-series classification

Time-Series Classification (TSC) has gained substantial importance in applications such as healthcare, finance, manufacturing, and human activity recognition. Training and evaluating a model on data with diverse distributions pose challenges across domains, but Unsupervised Domain Adaptation (UDA) proves efficient in mitigating these distribution discrepancies. Current TSC-UDA methods exploit statistical, adversarial, or contrastive learning strategies. Even though contrastive learning is an emerging UDA, relying solely on it proves inadequate in addressing the non-linear characteristics of the non-stationary multivariate temporal data. Kernelization enables better capturing of these non-linear relationships among the samples. Hence, an optimal kernel is derived by applying multiple kernel variants to samples and the maximum mean discrepancy metric is computed. The augmented samples in contrastive learning can result in overfitting, hence L1norm regularization is adopted to eliminate it. Therefore, we introduce a comprehensive approach Multikernel coNtrastivE doMain adaptatiON for tIme-series Classification (MNEMONIC) that embraces statistical feature alignment, contrastive learning, and regularization. Consequently, the model learns the inherent relationship among non-monotonic temporal data points and aligns the domains into a shared subspace, reducing the distribution differences. Extensive experimental evaluation is performed and the results are compared to the state-of-the-art with a macro-F1 score of 94.36%, 86.19%, 70.08%, and 63.07% for UCIHAR, HHAR, WISDM, and SSC datasets respectively.

Guided wave-based cross-scene interfacial debonding detection in reinforced concrete structures

The cross-scene problem within the field of guided wave-based interfacial debonding detection in steel-reinforced concrete structures is a significant challenge. This issue arises when the methods initially designed for laboratory-configured structures or numerical models are adapted to real-world scenarios. In response to this challenge, driven by the concept of domain adaptation, a novel method—the discriminative wavelet adversarial adaptation network (DWAAN)—is proposed to assess the health of analogous structures using diagnostic insights from a reference structure. The proposed method utilizes continuous wavelet transform to segregate signal components. A feature extraction network, designed with the convolutional neural network architecture, extracts domain confusion and damage-sensitive features through guided adversarial adaptive training in the wavelet domain. Both the adversarial and statistical alignment are employed during training for comprehensive optimization of data distribution discrepancies. The effectiveness of the proposed method is assessed in both the homogeneous and heterogeneous cross-scene interfacial debonding detection scenarios, revealing an average performance enhancement of 32.5% in distinct homogeneous applications and 26.6% in heterogeneous scenarios, when compared to the traditional supervised learning approach. In addition, the present method demonstrates robust convergence stability, evidenced by the minimized training fluctuations. The proposed method holds promising potential for practical applications in interfacial debonding detection in reinforced concrete structures while addressing the cross-scene challenge.

Locality Robust Domain Adaptation for cross-scene hyperspectral image classification

Domain adaptation (DA) has become a widely used technique for cross-scene hyperspectral image (HSI) classification. Most DA methods aim to learn a domain invariant subspace that reduce the domain discrepancy between source and target domains. However, some of them fail to explore the local manifold structure between different domains, while also neglecting the negative influences of abnormal features. To solve these problems, we propose a Locality Robust Domain Adaptation (LRDA) method for cross-domain data recognition. In LRDA, the statistical alignment is applied to reduce the domain-shift between the source and target domains. Then, LRDA combines the row-sparsity constraint and the discriminant regularization term to learn a robust projection matrix, while maintaining the discriminative capability of the matrix. Furthermore, a manifold regularization term is proposed to automatically learn the nearest neighbors and weights between two domains. The proposed LRDA not only reduces the discrepancy between two domains but also explores the local neighbor information between them. Experiment results on three HSI datasets illustrate that the proposed LRDA has better performance than other related methods.

Discriminative subspace embedded dynamic geometrical and statistical alignment based on pseudo-label correction for cross-domain bearing fault diagnosis

Many domain adaptation (DA) approaches have been developed to address the challenge of domain divergence in cross-domain fault diagnosis. However, most of them only attempt to align statistical distribution while neglecting geometric alignment between source and target data. Furthermore, the use of some unreliable pseudo-labels may cause geometrical and statistical distributions mismatching and interfere with the DA model generating correct pseudo-labels during the iterative learning. In this paper, we propose a new model called discriminative subspace embedded dynamic geometrical and statistical alignment based on pseudo-label correction (DSDGSA-PC) for bearing fault diagnosis. Firstly, discriminative subspace alignment is proposed to mitigate feature redundancy and divergence by generating aligned subspaces for two domains, while preserving class discriminative information and global structures of data. Then, DSDGSA-PC leverages the representer theorem and the principle of structural risk minimization to learn a domain-invariant classifier in the subspace, while minimizing statistical and geometrical shift by jointly optimizing dynamic graph embedding and dynamic weighted distribution alignment strategies. Finally, a novel pseudo-label correction mechanism is integrated into DSDGSA-PC to evaluate the credibility of pseudo-labels and rectify the unreliable ones during the iterations. The experimental results illustrate that DSDGSA-PC has higher transfer performance compared to several advanced methods on 24 transfer tasks.

WAD-X: Improving Zero-shot Cross-lingual Transfer via Adapter-based Word Alignment

Multilingual pre-trained language models (mPLMs) have achieved remarkable performance on zero-shot cross-lingual transfer learning. However, most mPLMs implicitly encourage cross-lingual alignment in pre-training stage, making it hard to capture accurate word alignment across languages. In this paper, we propose Word-align ADapters for Cross-lingual transfer (WAD-X) to explicitly align word representations of mPLMs via language-specific subspace. Taking a mPLM as the backbone model, WAD-X constructs subspace for each source-target language pair via adapters. The adapters use statistical alignment as the prior knowledge to guide word-level aligning in the corresponding bilingual semantic subspace. We evaluate our model across a set of target languages on three zero-shot cross-lingual transfer tasks: part-of-speech tagging (POS), dependency parsing (DP), and sentiment analysis (SA). Experimental results demonstrate that our proposed model improves zero-shot cross-lingual transfer on three benchmarks, with improvements of 2.19, 2.50, and 1.61 points in POS, DP, and SA tasks over strong baselines.

Research on blade statistics measurement registration planning technology

Blade has the characteristics of high machining quality of complex curved surface. If there is no benchmark before machining, it is impossible to judge whether the blades before machining are qualified. Therefore, it is necessary to analyze the blade measurement data. Due to the large measurement point error and disordered distribution, it is necessary to optimize the blade registration. Therefore, blade model registration and positioning is particularly important in blade shape detection and analysis. First, preregistration is carried out based on the six point optimization selection of the blade. After preregistration, the selection method of registration control point set based on theoretical model and statistical error is proposed, planning the registration datum point set on the blade model. The registration control point set is obtained through the registration operation between the measurement data and the registration reference point set. Finally, based on the stability and reliability of important sampling sensitivity and statistical error, obtain the probability density function of error normal distribution statistics samples and important samples. The selection of statistical control points and the rationality of the objective function were verified. The stability/reliability of the statistical alignment point selection is proved to be feasible. The statistical registration deviation is [0.015,0.026] mm, and the ICP registration deviation is [0.031,0.035] mm. The average deviation of statistics registration is about 0.013 mm smaller than the average deviation of ICP registration. The deviation of statistical sampling points is about 0.0214 mm, and that of traditional sampling points is about 0.0275 mm. The deviation of statistical sampling points is about 0.0061 mm smaller than that of traditional sampling points. It meets the requirements of rapid, high efficiency and high precision measurement for aeroengine blades.

When the Stars Align: A 5σ Concordance of Planetary Nebulae Major Axes in the Center of Our Galaxy

We report observations of a remarkable major axes alignment nearly parallel to the Galactic plane of 5σ significance for a subset of bulge “planetary nebulae” (PNe) that host, or are inferred to host, short-period binaries. Nearly all are bipolar. It is solely this specific PN population that accounts for the much weaker statistical alignments previously reported for the more general bulge PNe. It is clear evidence of a persistent, organized process acting on a measurable parameter at the heart of our galaxy over perhaps cosmologically significant periods of time for this very particular PN sample. Stable magnetic fields are currently the only plausible mechanism that could affect multiple binary star orbits as revealed by the observed major axes orientations of their eventual PNe. Examples are fed into the current bulge PN population at a rate determined by their formation history and mass range of their binary stellar progenitors.

Subspace and second-order statistical distribution alignment for cross-domain recognition of human hand motions

Subspace and second-order statistical distribution alignment for cross-domain recognition of human hand motions

Test-time bi-directional adaptation between image and model for robust segmentation

Background and objectiveDeep learning models often suffer from performance degradations when deployed in real clinical environments due to appearance shifts between training and testing images. Most extant methods use training-time adaptation, which almost require target domain samples in the training phase. However, these solutions are limited by the training process and cannot guarantee the accurate prediction of test samples with unforeseen appearance shifts. Further, it is impractical to collect target samples in advance. In this paper, we provide a general method of making existing segmentation models robust to samples with unknown appearance shifts when deployed in daily clinical practice. MethodsOur proposed test-time bi-directional adaptation framework combines two complementary strategies. First, our image-to-model (I2M) adaptation strategy adapts appearance-agnostic test images to the learned segmentation model using a novel plug-and-play statistical alignment style transfer module during testing. Second, our model-to-image (M2I) adaptation strategy adapts the learned segmentation model to test images with unknown appearance shifts. This strategy applies an augmented self-supervised learning module to fine-tune the learned model with proxy labels that it generates. This innovative procedure can be adaptively constrained using our novel proxy consistency criterion. This complementary I2M and M2I framework demonstrably achieves robust segmentation against unknown appearance shifts using existing deep-learning models. ResultsExtensive experiments on 10 datasets containing fetal ultrasound, chest X-ray, and retinal fundus images demonstrate that our proposed method achieves promising robustness and efficiency in segmenting images with unknown appearance shifts. ConclusionsTo address the appearance shift problem in clinically acquired medical images, we provide robust segmentation by using two complementary strategies. Our solution is general and amenable for deployment in clinical settings.

Ensemble Alignment Subspace Adaptation Method for Cross-Scene Classification

An ensemble alignment subspace adaptation method is proposed in this letter for the cross-scene classification. It can settle the problem of both foreign objects in the same spectrum and different spectrums. The algorithm combines the idea of ensemble learning with the domain adaptive (DA) algorithm. Considering the sample imbalance problem of the original data (OD), the source data (SD) is obtained by multiple random sampling of OD according to certain rules and used as input. Then, geometric alignment and statistical alignment of SD and target data (TD) are performed to build a communal subspace, followed by the classification of TD. The classification labels are finally ensembled by counting the multiple classification results with retaining valid information. This technique can reduce the uncertainty and randomness of generating subspace projections. The experimental results on two real datasets show that the proposed algorithm has a terrific accuracy improvement compared with the traditional machine learning and DA methods.

Statistical modeling: Assessing the anatomic variability of knee joint space width

Population-based knee joint space width (JSW) assessments are promising for the prevention and early diagnosis of osteoarthritis. This study aimed to establish the statistical shape and alignment model (SSAM) of knee joints for assessing anatomic variation in knee JSW in the healthy Chinese male population. CT scans of asymptomatic knee joints of healthy male participants (n = 107) were collected for manual segmentation to create mesh samples. The as-scanned positional error was reduced by a standard processing flow of deformable mesh registration. Principal component analysis (PCA) was performed to create a tibiofemoral SSAM that was trained on all mesh samples. The anatomic variability of the JSW in the healthy Chinese male population was then assessed using the SSAM with regression analysis and 3D analysis by color-coded mapping. Almost all PCA modes had a linear influence on the anatomic variation of the medial and lateral JSW. The JSW variability within the SSAM was mainly explained by mode 1 (45.1 % of variation), demonstrating that this mode had the greatest influence on JSW variation. 3D assessment of the JSW showed that the minimum medial JSW varied from 2.76 to 3.23 mm, and its site shifted a short distance on the medial tibial plateau. The root-mean-square fitting and generalization errors of the SSAM were below 1 mm. This study will benefit the design and optimization of prosthetic devices, and may be applicable to the prevention and early diagnosis of osteoarthritis.

A general approach to calculate the stiffness tensor of short‐fiber composites using the fabric tensor determined by X‐ray computed tomography

AbstractIn this work, a general method to predict the stiffness tensor of short‐fiber composites as a function of the fabric tensor (i.e. the tensor describing the architectural anisotropy of the micro‐structure), determined by X‐ray computed tomography, is presented. The proposed method does not depend on the type of fabric tensor used to characterize the symmetry properties of the composite. The experimental data used for method validation were obtained from thermoplastic composite matrices reinforced with short carbon and basalt fibers. The components of the stiffness tensor were calculated using the Mori‐Tanaka model and the Halpin‐Tsai equation for the stiffness of aligned short‐fiber composites, but other micromechanics models could be used as well. The statistical alignment of the fibers in each portion of the sample was measured by determining the principal axes of the fabric tensor. The average properties calculated on sub‐volumes provided moduli values in very good agreement with those determined experimentally. This approach is very general and easy to implement and does not need any numerical analysis tools.

Deep multi feature dynamic adversarial diagnosis approach of rotating machinery

Recent works show that knowledge transfer is an effective strategy to solve cross-domain diagnosis problems. The existing domain adaptation methods considering both global and local distribution between domains do not make the most of the knowledge learned by deep neural network, resulting in low diagnosis accuracy. To solve this problem, a deep multi feature dynamic adversarial diagnosis (DMDAD) method for transfer diagnosis of rotating machinery is presented in this paper. Firstly, the one-dimensional deep convolutional neural network is utilized as the feature encoder to learn the characteristics of vibration signals in different working conditions. The class prediction vector and feature vector are fused by the multilinear mapping. The fused features are conducted for the dynamic discrimination network for adversarial training. At the same time, considering the statistical alignment and adversarial alignment, the domain adaptation is finally realized. The experimental study demonstrates the effectiveness of the DMDAD.

Manifold embedded joint geometrical and statistical alignment for visual domain adaptation

Supervised learning algorithms like KNN assume that the training and the testing data come from the same source, and hence, have the same distribution. However, in real life, such instances are rare. Domain adaptation (DA) aims to model a classifier that is invariant to the different features of the training data. Domain invariant features are utilized to reduce distribution differences, ensuring an iterative improvement on the predicted target labels. Existing DA approaches either perform manifold subspace learning or align cross-domain distributions. At the same time, there exists a couple of imperative limitations: (1) degenerated feature transformation, wherein the seemingly domain invariant features are distorted during transformation (2) formulation of a robust objective function that must consider critical objectives in a unified framework to minimize deviations geometrically and statistically. To the best of our knowledge, none of the existing DA approaches address these two limitations simultaneously. Therefore, we propose a novel domain adaptation framework, called Manifold Embedded Joint Geometrical and Statistical Alignment (MEJGSA) for visual domain adaptation to address these limitations. MEJGSA first learns manifold features, and then formulates a robust objective function to reduce divergence between domains geometrically and statistically. We also consider more reliable classifiers to generate pseudo-labels to preserve objective functions such as conditional distributions and target domain discriminant information. Extensive experiments on two widely used cross-domain adaptation datasets demonstrate that MEJGSA exhibits significant improvements in classification accuracy as compared to cutting-edge shallow and deep DA approaches. The MATLAB code for our proposed method will be publicly available at 11https://github.com/rakesh1000/MEJGSA.

An enhanced Gerber statistic for portfolio optimization

The purpose of this letter is to introduce a modified version of the Gerber statistic resulting in the enhancement of its function as a measure of statistical co-movement. The modification centres around the redefinition of zones and thresholds, alongside the use of finite real-valued contributions rather than discrete binary counts in assigning value to individual co-movements. Arguments for the former are based on the statistical alignment of data series for the purposes of more effectively delineating concordant and discordant co-movement. Arguments for the latter are based on attributing greater weight to better information. Collectively this approach simultaneously incorporates the merits of Gerber-like thresholds and zoning, and conventional correlation-like co-movement measurement.

Image features influence reaction time

We aim to ask and answer an essential question " how quickly do we react after observing a displayed visual target?" To this end, we present psychophysical studies that characterize the remarkable disconnect between human saccadic behaviors and spatial visual acuity. Building on the results of our studies, we develop a perceptual model to predict temporal gaze behavior, particularly saccadic latency, as a function of the statistics of a displayed image. Specifically, we implement a neurologically-inspired probabilistic model that mimics the accumulation of confidence that leads to a perceptual decision. We validate our model with a series of objective measurements and user studies using an eye-tracked VR display. The results demonstrate that our model prediction is in statistical alignment with real-world human behavior. Further, we establish that many sub-threshold image modifications commonly introduced in graphics pipelines may significantly alter human reaction timing, even if the differences are visually undetectable. Finally, we show that our model can serve as a metric to predict and alter reaction latency of users in interactive computer graphics applications, thus may improve gaze-contingent rendering, design of virtual experiences, and player performance in e-sports. We illustrate this with two examples: estimating competition fairness in a video game with two different team colors, and tuning display viewing distance to minimize player reaction time.