Local Transformations Research Articles

A hybrid enhanced attention transformer network for medical ultrasound image segmentation

Ultrasound (US) technology is a widely utilized for clinical screening owing to its cost-effectiveness, painlessness, and convenience. However, the automatic segmentation of lesions or organs in US images remains to be challenging due to speckle artifacts, blurred boundaries, and low contrast. Recently, transformer-based methods have shown to be effective in long-range dependency, which is a good complement to Convolutional Neural Networks (CNN). In this article, we present a novel U-shape segmentation model based on a hybrid CNN-transformer structure that can effectively integrate CNN local features and transformer long-range contextual information of US images. To begin with, we design a coordinate residual block (CdRB) to encode the absolute position information of lesions. Further, we develop a channel enhanced self-attention-based transformer (ECAT) to help enhance the response of extracted global features. Finally, we adopt a comprehensive dual attention module (CDAM) to enhance skip connection features, which can learn feature correlations and capture more accurate edge features in US images. Results based on four public US datasets demonstrate that our method outperforms state-of-the-art segmentation methods, with 0.741 Dice on BUSI for breast lesion segmentation, 0.827 Dice on DDTI for thyroid lesion segmentation, 0.895 Dice on TN3k for thyroid lesion segmentation and 0.940 Dice on CAMUS for left ventricle segmentation. Furthermore, the robustness of our network is further demonstrated by an external validation dataset for breast lesion segmentation. In summary, our method showcases excellent adaptability and robustness in US image segmentation and can potentially be a general US segmentation tool.

Pyramid-based self-supervised learning for histopathological image classification

Large-scale labeled datasets are crucial for the success of supervised learning in medical imaging. However, annotating histopathological images is a time-consuming and labor-intensive task that requires highly trained professionals. To address this challenge, self-supervised learning (SSL) can be utilized to pre-train models on large amounts of unsupervised data and transfer the learned representations to various downstream tasks. In this study, we propose a self-supervised Pyramid-based Local Wavelet Transformer (PLWT) model for effectively extracting rich image representations. The PLWT model extracts both local and global features to pre-train a large number of unlabeled histopathology images in a self-supervised manner. Wavelet is used to replace average pooling in the downsampling of the multi-head attention, achieving a significant reduction in information loss during the transmission of image features. Additionally, we introduce a Local Squeeze-and-Excitation (Local SE) module in the feedforward network in combination with the inverse residual to capture local image information. We evaluate PLWT’s performance on three histopathological images and demonstrate the impact of pre-training. Our experiment results indicate that PLWT with self-supervised learning performs highly competitive when compared with other SSL methods, and the transferability of visual representations generated by SSL on domain-relevant histopathological images exceeds that of the supervised baseline trained on ImageNet.

Performance Assessment of Bias Correction Methods for Precipitation and Temperature from CMIP5 Model Simulation

Hydrological modeling relies on the inputs provided by General Circulation Model (GCM) data, as this allows researchers to investigate the effects of climate change on water resources. But there is high uncertainty in the climate projections with various ensembles and variables. Therefore, it is very important to carry out bias correction in order to analyze the impacts of climate change at a regional level. The performance evaluation of bias correction methods for precipitation, maximum temperature, and minimum temperature in the Upper Bhima sub-basin has been investigated. Four bias correction methods are applied for precipitation viz. linear scaling (LS), local intensity scaling (LOCI), power transformation (PT), and distribution mapping (DM). Three bias correction methods are applied for temperature viz. linear scaling (LS), variance scaling (VS), and distribution mapping (DM). The evaluation of the results from these bias correction methods is performed using the Kolmogorov–Smirnov non-parametric test. The results indicate that bias correction methods are useful in reducing biases in model-simulated data, which improves their reliability. The results of the distribution mapping bias correction method have been proven to be more effective for precipitation, maximum temperature, and minimum temperature data from CMIP5-simulated data.

Specifics of knowledge reproduction in the digital economy: Social aspect

Introduction. The formation of the digital economy, which began in the second half of the 20th century, posed a lot of different problems to humanity. Many of them are associated with the specifics of the digitalization of social production, which, along with human capital, becomes the main resource for development. At the same time, the local digital transformation of economic relations entails changes in their entire scope. First of all, such changes concern the process of reproduction of knowledge, or rather, the technology of its «translation» from a «simple» to an «expanded» state. Purpose setting. The logic of the article is built in the discourse of theorizing the process of cooperation between a machine (AI) and a human. Cooperation (technology, tool) leads to changes in the very nature of work and, accordingly, the transformation of education as its service subsystem. The purpose of the article is to consider the process of knowledge reproduction in the educational architecture of the digital economy. Methodology and methods of the study. The analysis of such meta-processes involves an interdisciplinary analysis, and a structural approach will allow interpreting the main stages of knowledge reproduction in historical discourse. Taking into account the considerable number of economic and socio-philosophical studies, the article necessarily used the latest developments of domestic and foreign scientists. Results. The authors proposed a working definition of the digital economy. It is considered as a modern, historically conditioned way of economic development of society, which is based on the intensive expanded reproduction of knowledge in the form of cooperation between a machine (AI) and a human. The purpose and meaning of this type of reproduction is the use of the subjectivity of an individual in the direction of increasing his creative principle. Creativity is understood as the ability to generate knowledge based not only on material (rational), but also on irrational (ideal) values and meanings. According to these theses, the transformation of all service architectures of social development and, first of all, education should be carried out. Conclusion. The development of intangible assets should be a defining task in the formation of human capital in the digital economy. In turn, the holistic development of the individual»s intelligence in all its modifications (primarily cognitive, emotional and digital), that do not contradict each other is the general task of the education system.

Study on the Damage Effects of Explosive Media Based on FEM-DEM Local Adaptive Transformation

Study on the Damage Effects of Explosive Media Based on FEM-DEM Local Adaptive Transformation

Cformer: An underwater image enhancement hybrid network combining convolution and transformer

AbstractUnderwater images are the most direct and effective ways to obtain underwater information. However, underwater images typically suffer from contrast reduction and colour distortion due to the absorption and scattering of water by light, which seriously limits the further development of underwater visual tasks. Recently, the convolutional neural network has been extensively applied in underwater image enhancement for its powerful local information extraction capabilities, but due to the locality of convolution operation, it cannot capture the global context well. Although the recently emerging Transformer can capture global context, it cannot model local correlations. Cformer is proposed, which is an Unet‐like hybrid network structure. First, a Depth Self‐Calibrated block is proposed to extract the local features of the image effectively. Second, a novel Cross‐Shaped Enhanced Window Transformer block is proposed. It captures long‐range pixel interactions while dramatically reducing the computational complexity of feature maps. Finally, the depth self‐calibrated block and the cross‐shaped enhanced window Transformer block are ingeniously fused to build a global–local Transformer module. Extensive ablation studies are performed on public underwater datasets to demonstrate the effectiveness of individual components in the network. The qualitative and quantitative comparisons indicate that Cformer achieves superior performance compared to other competitive models.

'Hold the course(s)!' - a qualitative interview study of the impact of pandemic-triggered contact restrictions on online instruction in community-based family medicine teaching.

The COVID-19 pandemic has been identified as a catalyst for the digitalization of medical education. Less is known about the specific impact of the pandemic on decentralized, community-based education, such as in General Practitioner practices. The aim of this study is to understand the impact of the digital transformation process, triggered by the COVID-19 pandemic. As, family medicine courses involve both university-based and community-based teaching, this study focuses the mode and quality of instruction and instructors in family medicine teaching. A qualitative interview study was conducted. The participants (N = 12) of a multi-perspective Quality Circle in family medicine teaching were interviewed twice: first, in 2019, about digitalization in family medicine teaching in Tübingen, Germany, not yet aware of the global changes and local transformation that would take place shortly thereafter. Second, in a follow-up interview in 2020 about the transition process and digitalization following the impact of contact restrictions during the pandemic. Grounded Theory was used as a qualitative research approach to analyze the complex processes surrounding this transformation. By analyzing the interviews with various stakeholders of community and university-based teaching, a model for the digital transformation process of family medicine teaching at the University of Tübingen in response to an external stimulus (the pandemic) was developed. It involves six chronological steps: "The calm before the storm," "The storm hits," "All hands on deck," "Adrift," "Reset course," and "The silver lining." This model seeks to understand the process of digital transformation and its impact on the teaching institution (medical faculty of the University of Tübingen, Institute for General Practice and Interprofessional Health Care) and instructors from an integrated perspective and thereby critically revisits prior concepts and opinions on the digitalization of medical teaching. Insights gained are presented as key messages.

BRST cohomology is Lie algebroid cohomology

In this paper we demonstrate that the exterior algebra of an Atiyah Lie algebroid generalizes the familiar notions of the physicist's BRST complex. To reach this conclusion, we develop a general picture of Lie algebroid isomorphisms as commutative diagrams between algebroids preserving the geometric structure encoded in their brackets. We illustrate that a necessary and sufficient condition for such a diagram to define a morphism of Lie algebroid brackets is that the two algebroids possess gauge-equivalent connections. This observation indicates that the aforementioned set of Lie algebroid isomorphisms should be regarded as equivalent to the set of local diffeomorphisms and gauge transformations. Moreover, a Lie algebroid isomorphism being a chain map in the exterior algebra sense ensures that isomorphic algebroids are cohomologically equivalent. The Atiyah Lie algebroids derived from principal bundles with common base manifolds and structure groups may therefore be divided into equivalence classes of isomorphic algebroids. Each equivalence class possesses a local representative which we refer to as the trivialized Lie algebroid, and we show that the exterior algebra of the trivialized algebroid gives rise to the BRST complex. We conclude by illustrating the usefulness of Lie algebroid cohomology in computing quantum anomalies, including applications to the chiral and Lorentz-Weyl (LW) anomalies. In particular, we pay close attention to the fact that the geometric intuition afforded by the Lie algebroid (which was absent in the naive BRST complex) provides hints of a deeper picture that simultaneously geometrizes the consistent and covariant forms of the anomaly. In the algebroid construction, the difference between the consistent and covariant anomalies is simply a different choice of basis.

Conformer: Local Features Coupling Global Representations for Recognition and Detection.

With convolution operations, Convolutional Neural Networks (CNNs) are good at extracting local features but experience difficulty to capture global representations. With cascaded self-attention modules, vision transformers can capture long-distance feature dependencies but unfortunately deteriorate local feature details. In this paper, we propose a hybrid network structure, termed Conformer, to take both advantages of convolution operations and self-attention mechanisms for enhanced representation learning. Conformer roots in feature coupling of CNN local features and transformer global representations under different resolutions in an interactive fashion. Conformer adopts a dual structure so that local details and global dependencies are retained to the maximum extent. We also propose a Conformer-based detector (ConformerDet), which learns to predict and refine object proposals, by performing region-level feature coupling in an augmented cross-attention fashion. Experiments on ImageNet and MS COCO datasets validate Conformer's superiority for visual recognition and object detection, demonstrating its potential to be a general backbone network. Code is available at https://github.com/pengzhiliang/Conformer.

Global Prior Transformer Network in Intelligent Borescope Inspection for Surface Damage Detection of Aeroengine Blade

Surface damage detection is vital for diagnosis and monitoring of aero-engine blade. At present, borescope inspection is the dominant technology. Several inspectors hold borescope to inspect the blades one by one through naked eyes on the apron. The inspection of turbine blades even requires drilling into narrow aero-engine tail nozzle. The manual visual inspection is high cost and low efficiency. To improve detection efficiency and economic benefit, we propose an intelligent borescope inspection method in this paper. Facing the problem of weak damage information caused by background noise and unsatisfactory illumination, local window Transformer network efficiently models pixel-to-pixel relations with the help of global self-attention mechanism, shifted window strategy is used to conduct information exchange. The capacity of global modeling is benefit for capturing detail damage outline. Besides, to learn label relations as prior and embed it into model, semantic information of different damages is aggregated by a two-layer graph convolution network. The global label graph network provides global prior by modeling label dependencies based on the samples in dataset. Finally, the image features and label features are fused to provide rich feature representation for mode recognition and damage localization. We validate the effectiveness of the proposed method on three datasets, including simulated blade, aluminum, and real blade datasets. The results demonstrate that the proposed method has superior performance with 84.9 mAP on simulated blade dataset and satisfactory visualization results on real blade dataset.

Gauge symmetries and the Higgs mechanism in Quantum Finance

By using the Hamiltonian formulation, we demonstrate that the Merton-Garman equation emerges naturally from the Black-Scholes equation after imposing invariance (symmetry) under local (gauge) transformations over changes in the stock price. This is the case because imposing gauge symmetry implies the appearance of an additional field, which corresponds to the stochastic volatility. The gauge symmetry then imposes some constraints over the free parameters of the Merton-Garman Hamiltonian. Finally, we analyze how the stochastic volatility gets massive dynamically via Higgs mechanism.

Robust Face Alignment via Inherent Relation Learning and Uncertainty Estimation.

Human tends to locate the facial landmarks with heavy occlusion by their relative position to the easily identified landmarks. The clue is defined as the landmark inherent relation while it is ignored by most existing methods. In this paper, we present Dynamic Sparse Local Patch Transformer (DSLPT), a novel face alignment framework for the inherent relation learning and uncertainty estimation. Unlike most existing methods that regress facial landmarks directly from global features, the DSLPT first generates a rough representation of each landmark from a local patch cropped from the feature map and then adaptively aggregates them by a case dependent inherent relation. Finally, the DSLPT predicts the coordinate and uncertainty of each landmark by regressing their probability distribution from the output features. Moreover, we introduce a coarse-to-fine framework to incorporate with DSLPT for an improved result. In the framework, the position and size of each patch are determined by the probability distribution of the corresponding landmark predicted in the previous stage. The dynamic patches will ensure a fine-grained landmark representation for inherent relation learning so that a rough prediction result can gradually converge to the target facial landmarks. We integrate the coarse-to-fine model into an end-to-end training pipeline and carry out experiments on the mainstream benchmarks. The results demonstrate that the DSLPT achieves state-of-the-art performance with much less computational complexity. The codes and models are available at https://github.com/Jiahao-UTS/DSLPT.

A massive gauge theory à la Utiyama

Utiyama’s method is a deductive approach of building gauge theories for semi-simple groups of local transformations, including the Abelian U(1) case, the non-Abelian SU(N) group, and the gravitational interaction. Gauge theories à la Utiyama typically predict a massless gauge potential. This work brings a mass generation mechanism and Utiyama’s method together thus giving mass to the interaction boson without breaking the gauge symmetry. Herein we devote our attention to the Abelian case. Two gauge potentials are introduced: a vetor field A μ and a scalar field B. The associated gauge-invariant field strengths F μ ν and G μ are built from Utiyama’s technique. Gauge invariance requirement upon the total Lagrangian (including matter fields and gauge fields) yields the conserved currents. Finally, we study the simplest type of Lagrangian involving the field strengths and obtain the related field equation. By imposing appropriate constraints on this particular example, Stueckelberg model is recovered.

The Local Unscented Transform Kalman Filter for the Weather Research and Forecasting Model

In this study, the local unscented transform Kalman filter (LUTKF) proposed in the previous study estimates the state of the Weather Research and Forecasting (WRF) model through local analysis. Real observations are assimilated to investigate the analysis performance of the WRF-LUTKF system. The WRF model as a regional numerical weather prediction (NWP) model is widely used to explain the atmospheric state for mesoscale meteorological fields, such as operational forecasting and atmospheric research applications. For the LUTKF based on the sigma-point Kalman filter (SPKF), the state of the nonlinear system is estimated by propagating ensemble members through the unscented transformation (UT) without making any linearization assumptions for nonlinear models. The main objective of this study is to examine the feasibility of mesoscale data assimilations for the LUTKF algorithm using the WRF model and real observations. Similar to the local ensemble transform Kalman filter (LETKF), by suppressing the impact of distant observations on model state variables through localization schemes, the LUTKF can eliminate spurious long-distance correlations in the background covariance, which are induced by the sampling error due to the finite ensemble size; therefore, the LUTKF used in the WRF-LUTKF system can efficiently execute the data assimilation with a small ensemble size. Data assimilation test results demonstrate that the LUTKF can provide reliable analysis performance in estimating the WRF model state with real observations. Experiments with various ensemble size show that the LETKF can provide better estimation results with a larger ensemble size, while the LUTKF can achieve accurate and reliable assimilation results even with a smaller ensemble size.

A new class of string transformations for compressed text indexing

Introduced about thirty years ago in the field of data compression, the Burrows-Wheeler Transform (BWT) is a string transformation that, besides being a booster of the performance of memoryless compressors, plays a fundamental role in the design of efficient self-indexing compressed data structures. Finding other string transformations with the same remarkable properties of BWT has been a challenge for many researchers for a long time. In this paper, we introduce a whole class of new string transformations, called local orderings-based transformations, which have all the “myriad virtues” of BWT. As a further result, we show that such new string transformations can be used for the construction of the recently introduced r-index, which makes them suitable also for highly repetitive collections. In this context, we consider the problem of finding, for a given string, the BWT variant that minimizes the number of runs in the transformed string.

Research on the Application of Membership System in Market Management: Evidence from China

Under the strong competition in the retail industry, large warehouse membership supermarkets have a stable and large consumer base in the market with strong competitiveness. This paper selects two case studies, Costco, which is the world's leading warehouse membership supermarket, and Freshippo, which has newly emerged in the Chinese market, for comparative analysis to explore the industrial structure behind the membership system in the development process and special innovations made by local membership supermarkets in China to satisfy consumers. The Chinese retail market is significantly different and requires localized transformation and innovation in order to attract local consumers to join. The paper also explores Freshippo's attempts to increase the brand's competitiveness in the face of the enormous convenience of online shopping in China, namely, the innovation of an online + offline model and the addition of several consumer benefits to attract more consumers to join the membership family. Through the integration and analysis of the development strategies of two membership supermarkets, we offer specific advice on the next development of membership supermarkets.

A simple and efficient graph Transformer architecture for molecular properties prediction

Graph Transformer architecture started to catch fire in molecular properties prediction due to its ability to represent complex interactions between all nodes. However, the self-attention mechanism in Transformer encoder part transforms the graph data into a fully-connected graph for graph representation learning, causing the loss of the original structural information of the graph. In this work, Local Transformer is proposed to keep the original graph structure and aggregate local information of nodes. In the model, a simple graph convolution is designed to replace the self-attention module, which reaches the state-of-the-art performance on ZINC dataset. Further, in order to rectify the problem that graph neural networks(GNNs) cannot capture long-range interactions of atoms, a novel end-to-end framework combined the GNN with Local and Global Transformer is proposed, which achieves good results on QM9 dataset.

Point cloud-based scene flow estimation on realistically deformable objects: A benchmark of deep learning-based methods

Flow estimation on 3D point clouds is a challenging problem in the field of computer vision, which has great significance in many areas, such as autonomous driving and human interaction applications. Within the last years, the field of motion analysis has made great progress. The evaluation of the existing approaches mostly focuses on scenarios where objects are affected by rigid transformations. However, in many application areas such as gesture recognition or pose tracking, the detection of shape changes is essential and breaking them down to local rigid transformations is accompanied by loss of information. One component of our contributions is that we specifically prepared existing datasets for scene flow estimation on deformable objects. Additionally, we benchmark existing methods and analyze their behavior on various subtasks. The results show that already close to 80% of correct correspondences can be found on synthetic hand data, while only around 50% are found on real hand data. Our experimental validation and analysis help to build an understanding of new possibilities in broader areas. Furthermore, they should help to inspire possible further research directions.

Equivalent quantum systems

We have studied quantum systems on finite-dimensional Hilbert spaces and found that all these systems are connected through local transformations. Actually, we have shown that these transformations give rise to a gauge group that connects the Hamiltonian operators associated with each quantum system. This bridge allows us to connect different quantum systems, in such a way that studying one of them allows to understand the other through a gauge transformation. Furthermore, we included the case where the Hamiltonian operator can be time-dependent. An application for this construction will be achieved in the theory of control quantum systems.

Discovering the ultralow thermal conductive A2B2O7-type high-entropy oxides through the hybrid knowledge-assisted data-driven machine learning

Lattice engineering and distortion have been considered one kind of effective strategies for discovering advanced materials. The instinct chemical flexibility of high-entropy oxides (HEOs) motivates/accelerates to tailor the target properties through phase transformations and lattice distortion. Here, a hybrid knowledge-assisted data-driven machine learning (ML) strategy is utilized to discover the A2B2O7-type HEOs with low thermal conductivity (κ) through 17 rare-earth (RE = Sc, Y, La–Lu) solutes optimized A-site. A designing routine integrating the ML and high throughput first principles has been proposed to predict the key physical parameter (KPPs) correlated to the targeted κ of advanced HEOs. Among the smart-designed 6188 (5RE0.2)2Zr2O7 HEOs, the best candidates are addressed and validated by the principles of severe lattice distortion and local phase transformation, which effectively reduce κ by the strong multi-phonon scattering and weak interatomic interactions. Particularly, (Sc0.2Y0.2La0.2Ce0.2Pr0.2)2Zr2O7 with predicted κ below 1.59 Wm−1 K−1 is selected to be verified, which matches well with the experimental κ = 1.69 Wm−1 K−1 at 300 K and could be further decreased to 0.14 Wm−1 K−1 at 1473 K. Moreover, the coupling effects of lattice vibrations and charges on heat transfer are revealed by the cross-validations of various models, indicating that the weak bonds with low electronegativity and few bonding charge density and the lattice distortion (r*) identified by cation radius ratio (rA/rB) should be the KPPs to decrease κ efficiently. This work supports an intelligent designing strategy with limited atomic and electronic KPPs to accelerate the development of advanced multi-component HEOs with properties/performance at multi-scales.