Basic Transformations Research Articles

Data Augmentation in Earth Observation: A Diffusion Model Approach

High-quality Earth Observation (EO) imagery is essential for accurate analysis and informed decision making across sectors. However, data scarcity caused by atmospheric conditions, seasonal variations, and limited geographical coverage hinders the effective application of Artificial Intelligence (AI) in EO. Traditional data augmentation techniques, which rely on basic parameterized image transformations, often fail to introduce sufficient diversity across key semantic axes. These axes include natural changes such as snow and floods, human impacts like urbanization and roads, and disasters such as wildfires and storms, which limits the accuracy of AI models in EO applications. To address this, we propose a four-stage data augmentation approach that integrates diffusion models to enhance semantic diversity. Our method employs meta-prompts for instruction generation, vision–language models for rich captioning, EO-specific diffusion model fine-tuning, and iterative data augmentation. Extensive experiments using four augmentation techniques demonstrate that our approach consistently outperforms established methods, generating semantically diverse EO images and improving AI model performance.

Efficient Multiclass Classification of Small Datasets Using a Novel Contrast-based Learning Approach

Introduction: Deep learning models often face challenges in achieving optimal accuracy when classifying multiclass datasets, particularly when the dataset size is limited. This study introduces Contrast Based Learning (CBL), a novel data augmentation technique designed to address data scarcity. Method: CBL innovatively concatenates multiple image and contrast learning to generate enriched datasets that exhibit a higher diversity of complex features. By focusing on the contrasts between various images, this method enhances the model's ability to learn nuanced features, thereby improving generalization and reducing overfitting. Results: Unlike traditional data augmentation methods, which rely on basic transformations, CBL dynamically concatenates images from different classes, creating complex inputs that provide the model with a more comprehensive training dataset. Experimental results show that CBL significantly improves classification accuracy and outperforms state-of-the-art methods across multiple small-scale multiclass datasets. Conclusion: The findings highlight the robustness of CBL in addressing data limitations, demonstrating its potential to advance the classification performance of deep learning models.

Application of Thermography and Convolutional Neural Network to Diagnose Mechanical Faults in Induction Motors and Gearbox Wear

Nowadays, induction motors and gearboxes play an important role in the industry due to the fact that they are indispensable tools that allow a large number of machines to operate. In this research, a diagnosis method is proposed for the detection of different faults in an electromechanical system through infrared thermography and a convolutional neural network (CNN). During the experiment, we tested different conditions in the motor and the gearbox. The induction motor was operated in four conditions, in a healthy state, with one broken bar, a damaged bearing, and misalignment, while the gearbox was operated in three conditions with healthy gears, 50% wear, and 75% wear. The motor failures and gear wear were induced by different machining operations. Data augmentation was then performed using basic transformations such as mirror image and brightness variation. Ablation tests were also carried out, and a convolutional neural network with a basic architecture was proposed; the performance indicators show a precision of 98.53%, accuracy of 98.54%, recall of 98.65%, and F1-Score of 98.55%. The system obtained confirms that through the use of infrared thermography and deep learning, it is possible to identify faults at different points of an electromechanical system.

Diagnostic Frameworks in Arabic Poetry

Poets have relied in poetry on metaphorical performance, the effectiveness of which begins with groups of semantic transformations. Rhetoricians often begin metaphor analysis by discussing diagnosis, as it represents the two sides of the basic transformation. It is an advanced stage and an important means of drawing the metaphorical image to discover the poet's inner world and explore his unknown worlds, and his inner feelings with what he possesses of symbolic and psychological rules. The concept of personification overlaps with the concept of metaphorical performance through a set of rhetorical analyses by highlighting the similar graphic relation between these analyses. Metaphorical performance aims to convey the information, knowledge and experiences to the recipient by a productive diagnostic structure within the framework of a successive graphic and rhetorical structure of interconnected images. Thus, every metaphorical performance has its diagnostic units and its harmony in a specific rhetorical context constitutes a rhetorical transformation, whether in poetry or prose. The reader of a metaphorical text becomes a party to the diagnostic text, as the diagnosis is represented in the unnatural language relations, and this matter makes us consider the metaphorical text an achievement of a rhetorical system and a material embodiment of it.

Positive discrimination of minority classes through data generation and distribution: A case study in olive disease classification

Deep learning models have achieved remarkable success in various tasks, especially in classification. This success is particularly evident in the precise classification of plant diseases, which is crucial for effective agricultural management. However, accurate classification faces challenges, particularly in data collection, where certain classes are underrepresented, namely the minority classes. This issue can significantly impact model performance. To tackle this challenge, this paper introduces a novel methodology that differs from existing approaches. We focus on addressing the issue of minority classes in image-based classification tasks, particularly for olive diseases. We employ data generation methods, including basic transformations, to produce augmented data and utilize Deep Convolutional Generative Adversarial Networks (DCGAN) to produce generated data. Next, we apply the Frechet Inception Distance (FID) to the generated dataset to select the highest-quality images. We then distribute varying percentages (25%, 50%, 75%, 100%) of this new data into the minority classes of the original dataset. Our data distribution strategies involve incorporating specific amounts of (1) augmented data, (2) generated data, and (3) a combination of both augmented and generated data to achieve target percentages (T.P) in the resulting dataset. Our experiments focus on classifying olive diseases into seven distinct categories using a pre-trained Convolutional Neural Network (CNN) architecture. We observe significant improvements in the model’s performance, particularly in the accurate classification of minority classes. This approach enhances diagnostic accuracy and optimizes data distribution, which is crucial for effectively addressing the challenges posed by minority classes.

Closed-loop identification technology for the whole life cycle of the fixed defective state of relay protection equipment

Abstract Due to the multiplicity of fixed-value defect states of relay protection equipment, it is difficult to guarantee the effect of identification directly through a single indicator. Therefore, this paper proposes research on closed-loop identification technology for the full life cycle of fixed-value defect states of relay protection equipment. The generalized transformation ratio of the relay protection circuit is analyzed based on the CT basic transformation ratio, cable voltage drop correction coefficient, filter attenuation correction coefficient, and ADC quantization error correction coefficient. The relationship between the transformation ratio and the fixed value state of the relay protection equipment enables the identification of defective states. In the test results, the design technology can effectively identify fixed-value defects at different stages and to varying degrees.

Two P, Ten P, White P, Red P: Mechanistic Exploration of the Oligomerization of Red Phosphorus from Diphosphorus with the Ab Initio Nanoreactor.

Phosphorus is critical to humans on many fronts, yet we do not have a mechanistic understanding of some of its most basic transformations and reactions─namely the oligomerization of white phosphorus to red. With heat or under ultraviolet (UV) exposure, it has been experimentally demonstrated that white phosphorus dissociates into diphosphorus units which readily form red phosphorus. However, the mechanism of this process is unknown. The ab initio nanoreactor approach was used to explore the potential energy surface of phosphorus clusters. Density functional theory and metadynamics simulations were used to characterize potential reaction pathways. A mechanism for oligomerization is proposed to take place via diphosphorus additions at π-bonds and weak σ-bonds through three membered ring intermediates. Downhill paths through P6 and P8 clusters eventually result in P10 clusters that can oligomerize into red phosphorus chains. The initial, rate limiting step for this process has an energy barrier of 24.2 kcal/mol.

CvFormer: Cross-view transFormers with pre-training for fMRI analysis of human brain

In recent years, functional magnetic resonance imaging (fMRI) has been widely utilized to diagnose neurological disease, by exploiting the region of interest (RoI) nodes as well as their connectivities in human brain. However, most of existing works only rely on either RoIs or connectivities, neglecting the potential for complementary information between them. To address this issue, we study how to discover the rich cross-view information in fMRI data of human brain. This paper presents a novel method for cross-view analysis of fMRI data of the human brain, called Cross-view transFormers (CvFormer). CvFormer employs RoI and connectivity encoder modules to generate two separate views of the human brain, represented as RoI and sub-connectivity tokens. Then, basic transformer modules can be used to process the RoI and sub-connectivity tokens, and cross-view modules integrate the complement information across two views. Furthermore, CvFormer uses a global token for each branch as a query to exchange information with other branches in cross-view modules, which only requires linear time for both computational and memory complexity instead of quadratic time. To enhance the robustness of the proposed CvFormer, we propose a two-stage strategy to train its parameters. To be specific, RoI and connectivity views can be firstly utilized as self-supervised information to pre-train the CvFormer by combining it with contrastive learning and then fused to finetune the CvFormer using label information. Experiment results on two public ABIDE and ADNI datasets can show clear improvements by the proposed CvFormer, which can validate its effectiveness and superiority.

Targeting MELK in tumor cells and tumor microenvironment: from function and mechanism to therapeutic application.

Maternal embryonic leucine zipper kinase (MELK), a member of the adenosine monophosphate-activated protein kinase (AMPK) protein family, has been reported to be involved in the regulation of many cellular events. The aberrant expression of MELK is associated with tumorigenesis and malignant progression of various tumors. Moreover, MELK plays an essential role in the regulation of tumor microenvironment (TME), which affects the function of immune cells and the responsiveness to immunotherapy. Currently, small molecule inhibitors targeting MELK have been developed and evaluated in clinical trials. A comprehensive understanding of MELK may provide clues and confidence for subsequent basic research and scientific transformation. In this review, we provide a comprehensive overview of the structural features, molecular biological functions, and critical roles of MELK in tumors and TME, as well as the targeted agents under development for the treatment of tumors and discuss the perspective for MELK-targeted therapies for tumors.

A comprehensive transformation-thermomechanical model on deformation history in directed energy deposition of high-speed steel

ABSTRACT Phase transformation is a crucial factor that determines the quality and deformation of components in laser-directed energy deposition (L-DED). Owing to the limitations of in situ observation methods, there is a lack of effective means to study the complete phase transformation and their impacts on deformation during the deposition process. In this study, multilayer heterogeneous deposition was investigated with the high-carbon high-speed steel by modelling the coupled phase transformation-thermomechanical physics. The methodology was proposed to identify the specific phase transformations by iteratively comparing the whole deformation history between digital image correlation (DIC) measurements and simulations. Besides the thermo-mechanical coupling effect and basic volumetric phase transformation, carbon partitioning between precipitated carbides and matrix was manifested, which also resulted in a rise of the Ms point by 208.4 °C and a reduction of the transformation rate by 73.2%. Additionally, an increase of 13.8% in energy absorption could be attributed to the surface oxidation of the deposited layer. The model predictions exhibited good consistency with the DIC in-situ measurement data, indicating that the deformation history contained sufficient information on the phase transformations. Based on the proposed transformation-thermomechanical coupling model, it helps to understand the complete phase transformation during deposition.

Spatial entropy as an inductive bias for vision transformers

Recent work on Vision Transformers (VTs) showed that introducing a local inductive bias in the VT architecture helps reducing the number of samples necessary for training. However, the architecture modifications lead to a loss of generality of the Transformer backbone, partially contradicting the push towards the development of uniform architectures, shared, e.g., by both the Computer Vision and the Natural Language Processing areas. In this work, we propose a different and complementary direction, in which a local bias is introduced using an auxiliary self-supervised task, performed jointly with standard supervised training. Specifically, we exploit the observation that the attention maps of VTs, when trained with self-supervision, can contain a semantic segmentation structure which does not spontaneously emerge when training is supervised. Thus, we explicitly encourage the emergence of this spatial clustering as a form of training regularization. In more detail, we exploit the assumption that, in a given image, objects usually correspond to few connected regions, and we propose a spatial formulation of the information entropy to quantify this object-based inductive bias. By minimizing the proposed spatial entropy, we include an additional self-supervised signal during training. Using extensive experiments, we show that the proposed regularization leads to equivalent or better results than other VT proposals which include a local bias by changing the basic Transformer architecture, and it can drastically boost the VT final accuracy when using small-medium training sets. The code is available at https://github.com/helia95/SAR.

Applying a novel generalized model of sedimentation to the design of settlers and thickeners: Integration and extension of the classical procedures

A new generalized procedure previously introduced and justified by the authors, based on the simultaneous integration of the macroscopic and microscopic differential mass balances by introducing the concept of separation units as an extension of the transfer unit concept was applied here to sedimentation. This similarity leads to a broader analogy with basic transfer and transformation unit operations, at this step focused on the formulation and contrast of a generalized form of the so-called design or sizing equation. This model is comparatively applied under restricted flow Conditions-Vertical Flow Settlers (CSTS) - with the classical methods and extended to the design of horizontal flow settlers (PFS). That opens the door to the sizing of settlers and thickeners in the presence of deviations from these ideal flow models and with other operating configurations. Such a contribution to a desirable generalization and homologation is according to Chemical Engineering methodology specifically in the frequently empirical or casuistic treatment of unit operations with particulate solids.

ENVIRONMENTAL ASPECTS IN THE CONTENT OF HISTORICAL EDUCATION

The article analyses the processes of content transformations in the context of educational reform: the implementation of the New Ukrainian School concept and higher education, which outline the key issues of optimising the structure and content of the historical educational field of knowledge. The basis for these processes is the modernised State Standard of Basic Secondary Education and transformations in the content of professional training in higher education. The authors note that certain content transformations of the history curricula are underway, filled with new cross-cutting lines that include the issue of environmental education, the development of knowledge competencies in environmental protection and issues in the «human-nature» problem field. A special place in this context belongs to the development of cross-cutting knowledge of the history and impact of human activity on nature, land, rivers and seas, flora and fauna, climate, etc. The authors describe the main components of the Concept of Environmental Education in Ukraine (2001) and highlight its principles: comprehensiveness, continuity, and dissemination among the population, taking into account individual professional interests, incentives, and peculiarities of social and territorial groups to preserve the national natural landscape. Modern environmental/ «green» education is an ongoing and meaningful process of forming an environmental outlook, environmental awareness and culture in all segments of the population, social groups and society as a whole. It is the process of mastering a range of knowledge (historical, mathematical, natural, etc.) about the laws of functioning, life of all living things, ecological systems and the role of humans in preserving the natural environment; the process of individual environmental education and training, mastering professional knowledge and skills necessary for environmental protection and preservation of the fundamentals of natural landscapes. The purpose of the article is to study the content features of historical education through the prism of the ecological worldview of man, its genesis in understanding his active influence on nature in the course of history. The authors come to the conclusion that it is important to introduce environmental education into the content of all disciplines of secondary and higher education.

Probing the limit of hydrologic predictability with the Transformer network

For a number of years since their introduction to hydrology, recurrent neural networks like long short-term memory (LSTM) networks have proven remarkably difficult to surpass in terms of daily hydrograph metrics on community-shared benchmarks. Outside of hydrology, Transformers have now become the model of choice for sequential prediction tasks, making it a curious architecture to investigate for application to hydrology. Here, we first show that a vanilla (basic) Transformer architecture is not competitive against LSTM on the widely benchmarked CAMELS streamflow dataset, and lagged especially prominently for the high-flow metrics, perhaps due to the lack of memory mechanisms. However, a recurrence-free variant of the Transformer model obtained mixed comparisons with LSTM, producing very slightly higher Kling-Gupta efficiency coefficients (KGE), along with other metrics. The lack of advantages for the vanilla Transformer network is linked to the nature of hydrologic processes. Additionally, similar to LSTM, the Transformer can also merge multiple meteorological forcing datasets to improve model performance. Therefore, the modified Transformer represents a rare competitive architecture to LSTM in rigorous benchmarks. Valuable lessons were learned: (1) the basic Transformer architecture is not suitable for hydrologic modeling; (2) the recurrence-free modification is beneficial, so future work should continue to test such modifications; and (3) the performance of state-of-the-art models may be close to the prediction limits of the dataset. As a non-recurrent model, the Transformer may bear scale advantages for learning from bigger datasets and storing knowledge. This work lays the groundwork for future explorations into pretraining models, serving as a foundational benchmark that underscores the potential benefits in hydrology.

Structural pathway for nucleation and growth of topologically close-packed phase from parent hexagonal crystal

The solid diffusive phase transformation involving the nucleation and growth of new phase particles is universal and frequently employed but has not yet been fully understood at the atomic level. Here, our first-principles calculations reveal a structural formation pathway of a series of topologically close-packed (TCP) phases within the hexagonally close-packed (hcp) matrix. The results show that the nucleation follows a nonclassical nucleation process, and apart from atomic diffusion, the entire hcp→TCP transformation only involves shuffle-based displacements, with a specific 3-layer unstable hcp-ordering serving as the basic structural transformation unit (BSTU). The thickening of plate-like TCP phases relies on the formation of these unstable hcp-orderings at their coherent TCP/matrix interface to nucleate a ledge, but the ledge lacks the dislocation characteristics which is commonly considered in the conventional view. Additionally, the atomic structure of the critical nucleus for the Mg2Ca and MgZn2 Laves phases was predicted in terms of Classical Nucleation Theory (CNT). The formation of polytypes and off-stoichiometry in TCP precipitates is found to be related to the nonclassical nucleation behavior. These insights contribute to a deeper understanding of solid diffusive phase transformations at the atomic level and provide a foundation for investigating other technologically significant transformations.

Research progress in endometriosis-associated ovarian cancer.

Endometriosis-associated ovarian cancer (EAOC) is a unique subtype of ovarian malignant tumor originating from endometriosis (EMS) malignant transformation, which has gradually become one of the hot topics in clinical and basic research in recent years. According to clinicopathological and epidemiological findings, precancerous lesions of ovarian clear cell carcinoma (OCCC) and ovarian endometrioid carcinoma (OEC) are considered as EMS. Given the large number of patients with endometriosis and its long time window for malignant transformation, sufficient attention should be paid to EAOC. At present, the pathogenesis of EAOC has not been clarified, no reliable biomarkers have been found in the diagnosis, and there is still a lack of basis and targets for stratified management and precise treatment in the treatment. At the same time, due to the long medical history of patients, the fast growth rate of cancer cells, and the possibility of eliminating the earliest endometriosis-associated ovarian cancer, it is difficult to find the corresponding histological evidence. As a result, few patients are finally diagnosed with EAOC, which increases the difficulty of in-depth study of EAOC. This article reviews the epidemiology, pathogenesis, risk factors, clinical diagnosis, new treatment strategies and prognosis of endometriosis-associated ovarian cancer, and prospects the future direction of basic research and clinical transformation, in order to achieve stratified management and personalized treatment of ovarian cancer patients.

Research on Chinese-English Patent Machine Translation Based on Fusion Strategy Model

In order to improve the translation quality of the machine translation model in the field of patent text, this paper proposes a patent knowledge fusion strategy. On the one hand, it integrates the structural knowledge, special phrases and professional terms in the patent text into the translation model as embedded comments, so as to improve its translation effect. On the other hand, based on XLNet, the basic translation model of Transformer is improved, and the translation performance of the model is further improved on the basis of integrating professional knowledge. The simulation results show that compared with the basic Transformer model and the Transformer model that only introduces fusion strategy, BLEU values of the proposed machine translation model combining fusion strategy and improved Transformer for the Chinese-English and English-Chinese translation tasks has increased by 3.77, 1.79, and 2.17, 0.46, respectively. It has a significant improvement in improving the quality of patent literature translation and is worthy of further research and promotion.

FGeo-SSS: A Search-Based Symbolic Solver for Human-like Automated Geometric Reasoning

Geometric problem solving (GPS) has always been a long-standing challenge in the fields of automated reasoning. Its problem representation and solution process embody rich symmetry. This paper is the second in a series of our works. Based on the Geometry Formalization Theory and the FormalGeo geometric formal system, we have developed the Formal Geometric Problem Solver (FGPS) in Python 3.10, which can serve as an interactive assistant or as an automated problem solver. FGPS is capable of executing geometric predicate logic and performing relational reasoning and algebraic computation, ultimately achieving readable, traceable, and verifiable automated solutions for geometric problems. We observed that symmetry phenomena exist at various levels within FGPS and utilized these symmetries to further refine the system’s design. FGPS employs symbols to represent geometric shapes and transforms various geometric patterns into a set of symbolic operation rules. This maintains symmetry in basic transformations, shape constructions, and the application of theorems. Moreover, we also have annotated the formalgeo7k dataset, which contains 6981 geometry problems with detailed formal language descriptions and solutions. Experiments on formalgeo7k validate the correctness and utility of the FGPS. The forward search method with random strategy achieved a 39.71% problem-solving success rate.

Decomposition of strategic goals and objectives of the Samara region dental

Aim to carry out a scientifically based decomposition of the strategic goals and objectives of the dental service of the Samara region, to determine the necessary basic transformations and opportunities for standardization of the industry model of the separation of powers and responsibilities at various hierarchical levels of management.
 Material and methods. When preparing the article, the conceptual provisions of the National Project, Federal projects, materials of reports on the activities of the dental service of the Samara region in the period 2020 2022 were used. The decomposition was carried out using the main tools of strategic, programtarget and process management; requirements of TQM and PDCA; principles of management by goals, building a balanced system of indicators and key performance indicators.
 Results. The practical task of decomposition of strategic goals and objectives of the development of the dental service of the Samara region has been solved, taking into account the strategic goals of the healthcare of the Russian Federation, a multidimensional decomposition of strategic goals and objectives of the dental service of the region has been carried out; a list of necessary basic transformations with the separation of powers and responsibilities has been determined.

Intelligent Program Correction and Evaluation System

The growing popularity of computing applications has sparked the interest of students in computer programming languages. Minor mistakes are prevalent while writing small code blocks due to the coder’s lack of knowledge and carelessness. Instead of merely providing syntax warnings, it would be better to offer developers with an Integrated Development Environment (IDE) that can automatically correct short code blocks containing mistakes. This makes code composition easier and more time-efficient, thus improving the efficacy of large-scale development environments. Because Python is becoming more popular, the goal of our study is to enhance the efficiency of writing Python code by offering an automatic code-correcting approach. Furthermore, automatic program evaluation has been performed to assist in the debugging of small code blocks, which will ultimately be employed in the creation of real-time computer applications. The proposed technique would be useful for new learners who wish to create small Python code blocks for ease of writing and debugging on online platforms (like edX, Coursera, and Udacity). One of the major contributions of the project is to create an erroneous dataset of Python coding that contains all potential forms of probable syntax errors. The dataset induces variety through the use of multiple coding templates and is used to train deep learning models. We used the state-of-the-art text-to-text T5 transformer network model to automatically repair and evaluate the incorrect code. The outcomes of auto-correction are examined using the ROUGE and BLEU scores, as well as accuracy. The model corrects Python code with single, double, and the multiple number of errors with greater than 80% accuracy. Similarly, the performance of the basic T5 transformer network for program auto-evaluation with and without mistakes has been examined, and the model achieves greater than 65% accuracy in both cases. The proposed T5 base transformer outperforms the SOTA auto-correction models in terms of accuracy, according to a comparison study of the proposed method with the earlier techniques for auto-correcting codes.