Dependent Type Research Articles

Моделирование ожидаемого предотвращенного ущерба в энергетике. Часть 1. Детерминированный подход

Inspection control and supervisory activities are one of the important mechanisms of proactive industrial safety. Corporate control ensures early detection of security violations and their prompt elimination to prevent accidents. The expected prevented damage at hazardous production facilities is considered one of the efficiency factors of inspection activities. In order to assess the expected prevented damage (direct or indirect), the number of expected incidents and damage for each event must be calculated. An approach to the assessment of the number of prevented events based on the pyramid of incidents (Heinrich’s pyramid, Berd’s pyramid) by supplementing it with the level of detected violations has been proposed. The number of prevented incidents can be assessed within the framework of deterministic and scholastic approaches. The study focuses mainly on building and analysis of the ratio of the number of incidents between different levels of the pyramid within the framework of the deterministic approach. The accident statistics at oil and gas industry facilities has been analyzed; Heinrich’s hypothesis regarding the type of dependence between the pyramid levels has been tested; and coefficients of the injury rate pyramid have been determined. To find the final coefficients of the improved pyramid of an oil and gas company, the method of coefficient calculation with accumulations for each year has been chosen. It has been demonstrated that traditional values of coefficients of Heinrich’s pyramid and those obtained as a result of calculations do not comply, which can be explained by the significant difference in industrial technologies and efficiency of occupational safety measures as of today and at the moment of initial building of Heinrich’s pyramid. Based on the regression analysis and other forecasting methods (the Holt method, the Brown exponential smoothing method, the ARIMA model, and the moving average extrapolation method), modeling and forecasting injury rate factors have been conducted. The comparative analysis of methods to forecast the number of accidents has been carried out. It has been shown that it is advisable to apply the linear regression model to predict injury rates.

MZ세대 대학생들의 스트레스 대처방식에 대한 인식

Objectives The purpose of this study is to categorize how university students of the MZ generation cope with stress and analyze their characteristics. Methods Q methodology was used to confirm the subjective perception of MZ generation college students' stress response methods. At this time, the Q methodology aims to objectively measure human subjective aspects such as attitudes, beliefs, and values, and this methodology was used because in-depth opinions could be studied. Results Stress coping methods were categorized into five. The first type is a ‘considerate type’ that finds the cause of stress and carefully analyzes it to relieve stress. The second type is a ‘lone type’ who takes time alone to avoid people and tries to solve it. The third type is a ‘dependent type’ that tries to relieve stress with the people around him. The fourth type is an ‘immersive type’ that creates a new situation and copes with stress situations while immersed in other things. The fifth type is a ‘propulsion type’ that focuses on solving problems, not identifying problems. Conclusions First, it is necessary to connect various programs inside and outside the school to relieve stress for MZ generation college students who fall under the ‘considerate type’. Second, the ‘lone type’ first needs a program that helps them develop self-esteem, self-resilience, and self-efficacy. Third, rather than introducing a specific program, the ‘dependent type’ needs to provide an environment where they can communicate and interact smoothly with people around them. Fourth, since ‘immersive’ college students are the type to immerse themselves in hobbies or sports activities, it is necessary to provide conditions for various hobbies and sports activities. Fifth, ‘promoting’ college students should be provided with all possible information, activities, and programs to relieve stress to create conditions for them to promote themselves.

Deriving Dependently-Typed OOP from First Principles

The expression problem describes how most types can easily be extended with new ways to produce the type or new ways to consume the type, but not both. When abstract syntax trees are defined as an algebraic data type, for example, they can easily be extended with new consumers, such as print or eval , but adding a new constructor requires the modification of all existing pattern matches. The expression problem is one way to elucidate the difference between functional or data-oriented programs (easily extendable by new consumers) and object-oriented programs (easily extendable by new producers). This difference between programs which are extensible by new producers or new consumers also exists for dependently typed programming, but with one core difference: Dependently-typed programming almost exclusively follows the functional programming model and not the object-oriented model, which leaves an interesting space in the programming language landscape unexplored. In this paper, we explore the field of dependently-typed object-oriented programming by deriving it from first principles using the principle of duality. That is, we do not extend an existing object-oriented formalism with dependent types in an ad-hoc fashion, but instead start from a familiar data-oriented language and derive its dual fragment by the systematic use of defunctionalization and refunctionalization. Our central contribution is a dependently typed calculus which contains two dual language fragments. We provide type- and semantics-preserving transformations between these two language fragments: defunctionalization and refunctionalization. We have implemented this language and these transformations and use this implementation to explain the various ways in which constructions in dependently typed programming can be explained as special instances of the general phenomenon of duality.

Inductive Diagrams for Causal Reasoning

The Lamport diagram is a pervasive and intuitive tool for informal reasoning about “happens-before” relationships in a concurrent system. However, traditional axiomatic formalizations of Lamport diagrams can be painful to work with in a mechanized setting like Agda. We propose an alternative, inductive formalization — the causal separation diagram (CSD) — that takes inspiration from string diagrams and concurrent separation logic, but enjoys a graphical syntax similar to Lamport diagrams. Critically, CSDs are based on the idea that causal relationships between events are witnessed by the paths that information follows between them. To that end, we model “happens-before” as a dependent type of paths between events. The inductive formulation of CSDs enables their interpretation into a variety of semantic domains. We demonstrate the interpretability of CSDs with a case study on properties of logical clocks , widely-used mechanisms for reifying causal relationships as data. We carry out this study by implementing a series of interpreters for CSDs, culminating in a generic proof of Lamport’s clock condition that is parametric in a choice of clock. We instantiate this proof on Lamport’s scalar clock, on Mattern’s vector clock, and on the matrix clocks of Raynal et al. and of Wuu and Bernstein, yielding verified implementations of each. The CSD formalism and our case study are mechanized in the Agda proof assistant.

Group-wise monitoring of multivariate data with missing values

In this article, we aim to detect the changes in the process that generates multivariate data by monitoring their mean shift. To do this, we utilize a graphical tool known as a multivariate control chart. However, monitoring the mean of multivariate data poses two challenges: the grouped structure of individual observations and the presence of missing values. In this research, we introduce a novel method called HTC for monitoring group-wise multivariate data that includes missing values. HTC offers several advantages over existing methods. First, it is applicable to various types of dependence among individual observations within a group. Second, it provides a unique upper control limit (UCL) regardless of the missing data pattern. Lastly, HTC is computationally more efficient compared to resampling-based techniques. We conduct comprehensive numerical studies to evaluate the performance of the HTC method and compare it with the existing group-wise monitoring method, referred to as HTM. Compared to HTM, HTC achieves a higher true positive rate (TPR) while effectively controlling the in-control false alarm rate ( FAR 0 ) at a pre-determined level across various settings considered in our study. To illustrate its effectiveness, we applied HTC to monitoring multivariate environmental data collected from the manufacturing process of a semiconductor company in Korea.

Coloniality dressed in green: in its current form, climate finance risks becoming a new tool for colonial rule

As the impacts of the climate crisis are being increasingly felt, a critical part of the solution is said to be the issue of mobilising climate finance. Particularly for the Global South, climate finance is crucial for sustainable development; to simultaneously meet the challenges of the climate crisis while addressing issues around economic development, health, poverty and beyond. Yet at present, Southern progress is being held back by ongoing and evolving patterns of Northern neo-coloniality, including through finance and debt relationships. In a context where the mainstream approach to mobilising climate finance – centring private finance, derisked by the state – reflects the dominance of US-style market-based finance, climate finance in this form simply risks becoming a new mechanism by which Southern countries are exposed to new types of subordination and dependence. Instead, structural changes and policy space is required for the Global South to break away from Northern financial dependence. The Bridgetown Agenda and calls for a new Bretton Woods moment are important steps in this direction. However, in addition such countries need the ability to develop financial institutions and regulatory structures that can simultaneously direct credit towards priority areas, regulate capital flows, and develop infrastructure that is democratically owned and oriented towards the needs of public and environment.

The credibility of external audits of management systems – the audited enterprises’ perspective

PurposeThis paper deals with external audits, which are now commonly used in many industries (e.g. food, automotive and electrical). This study aims to assess whether a given organization meets the specific criteria. If the audit ends with a positive result, information about it is provided to selected interested parties, e.g. clients or contractors. Credibility is pivotal in adding value for all interested parties within the audit processes. This study seeks the factors which, in the opinion of the audited enterprises, have the most decisive impact on the credibility of external audits.Design/methodology/approachIn keeping with the extant literature, research questions were developed regarding the factors influencing the credibility assessment of external audits. Data collected from 100 companies in the Polish food sector were used to construct the model and carry out statistical analyses. Linear regression analyses were also applied to determine the key factors influencing the credibility of audits.FindingsThis study is part of the research trend on the rationality of external audits and certification of quality management systems. This paper identifies nine main factors shaping the credibility of external audits. Two of them have the most decisive influence on credibility. The first one is the professional audit method (procedure). The second factor is the auditor’s knowledge of the specificity of the audited area.Research limitations/implicationsThis study did not consider the impact that the image/credibility of the organization represented by the auditors may have on the reliability of audits. This is one of the fundamental limitations that should be considered when analyzing the obtained results. To recognize this type of dependence, additional research should be carried out. Another limitation is that the research covers the food industry only. It would be interesting to know the situation in other types of industries.Practical implicationsThis paper looks at the possibility of increasing the added value for audited enterprises. The proposed model can be used by managers of organizations conducting external audits and auditors to effectively use resources for process improvement, influencing the maximization of credibility of activities in the area of conformity assessment.Originality/valueThe originality of this study lies in adopting the perspective of audited enterprises in assessing the credibility of audits. To the best of the authors’ knowledge, this is the first study that adopts this approach. This paper contributes to the literature, particularly to better understand audited enterprises’ behavior (trust in audit results, satisfaction with audits, etc.).

Abstract ND12: Discovery of PF-07220060, a potent and selective CDK4 inhibitor

Abstract PF-07220060 was developed based on the clinical need to improve efficacy and safety outcomes in patients treated with currently approved dual CDK4/6 inhibitors, including Palbociclib. Although all dual CDK4/6 inhibitors increase response rates and prolong disease control, they show common dose-limiting hematologic toxicities, foremost neutropenia. This poses challenges to provide these agents at doses necessary to extinguish cell cycling in tumors. Human ER+ breast cancer cells are highly dependent on CDK4 but not CDK6. By contrast, CDK4 is dispensable for human bone marrow derived cells, or cells enriched for neutrophil precursors or hematopoietic stem cells. This is primarily due to the compensatory role of CDK6 in these compartments. These observations prompted us to develop PF-07220060, an aminopyrimidine-based inhibitor, endowed with 20-fold and 4-fold increased selectivity for CDK4 over CDK6 when compared to Palbociclib and Abemaciclib/Ribociclib, respectively. CDK4-over-CDK6 selectivity is explained by G-loop conformational differences between the two kinases, with the G-loop of CDK4 being more flexible, assuming a more open conformation when bound to CDK4 selective versus non-selective inhibitors. PF-07220060 reduces dose-limiting neutropenia in beagle dogs by 15-fold when compared with Palbociclib, consistent with early clinical data obtained with PF-07220060. Higher tolerated plasma concentrations of PF-07220060 versus dual CDK4/6 inhibitors enable increased inhibition of CDK4 kinase activity in tumors. This translates to improved tumor growth inhibition across CDK4-driven models of human breast and prostate cancer at clinically relevant doses. Our work establishes proof-of-concept for CDK4 selective inhibition and introduces a best-in-class CDK4 inhibitor for treatment of HR+ HER2- breast cancer and other CDK4 dependent tumor types. Citation Format: Lars Anders, Gary Gallego. Discovery of PF-07220060, a potent and selective CDK4 inhibitor [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(7_Suppl):Abstract nr ND12.

Superparamagnetic effects in the linear magnetic response of polydisperse ensembles of nanoparticles suspended in a fluid.

Within a kinetic theory, the linear magnetic response of uniaxial single-domain particles suspended in a fluid is analyzed. The main qualitatively different types of frequency dependence of the longitudinal dynamic magnetic susceptibility of such particles are described. It is shown that superparamagnetic (related to orientation thermal fluctuations of the magnetic moment inside a particle) peculiarities of the response of a particle to a probing magnetic field are not fully determined by the ratio of anisotropy energy to thermal energy when a stationary bias field is applied. For a case where the indicated ratio is much greater than one, a simple approximate expression for the dynamic magnetic susceptibility of a particle is proposed. The developed approach is extended to polydisperse suspensions of noninteracting uniaxial nanoparticles. It is shown that polydispersity does not vanish away specific superparamagnetic features in the dynamic magnetic response of such systems. Quantitative estimates of the corresponding effects are performed in different frequency ranges of the applied field. It is demonstrated that under certain restrictions on the disperse composition of a suspension, the internal diffusion of the magnetic moment can lead to a splitting of the absorption spectrum of the system. The significant role of the bias field is revealed. In particular, it can cause an additional absorption maximum provided the particle-size distribution meets the outlined condition. Also, it enables one to assess how important it is to take into account superparamagnetism of particles: the effect of the biasing is stronger for particles with smaller anisotropy and thereby more pronounced superparamagnetic properties. A qualitative agreement of some of the inferences with the experimental data is briefly discussed.

Dependence on Tail Copula

In real-world scenarios, we encounter non-exchangeable dependence structures. Our primary focus is on identifying and quantifying non-exchangeability in the tails of joint distributions. The findings and methodologies presented in this study are particularly valuable for modeling bivariate dependence, especially in fields where understanding dependence patterns in the tails is crucial, such as quantitative finance, quantitative risk management, and econometrics. To grasp the intricate relationship between the strength of dependence and various types of margins, we explore three fundamental tail behavior patterns for univariate margins. Capitalizing on the probabilistic features of tail non-exchangeability structures, we introduce graphical techniques and statistical tests designed for analyzing data that may manifest non-exchangeability in the joint tail. The effectiveness of the proposed approaches is illustrated through a simulation study and a practical example.

SAGL: A self-attention-based graph learning framework for predicting survival of colorectal cancer patients

Background and Objective: Colorectal cancer (CRC) is one of the most commonly diagnosed cancers worldwide. The accurate survival prediction for CRC patients plays a significant role in the formulation of treatment strategies. Recently, machine learning and deep learning approaches have been increasingly applied in cancer survival prediction. However, most existing methods inadequately represent and leverage the dependencies among features and fail to sufficiently mine and utilize the comorbidity patterns of CRC. To address these issues, we propose a self-attention-based graph learning (SAGL) framework to improve the postoperative cancer-specific survival prediction for CRC patients.Methods: We present a novel method for constructing dependency graph (DG) to reflect two types of dependencies including comorbidity-comorbidity dependencies and the dependencies between features related to patient characteristics and cancer treatments. This graph is subsequently refined by a disease comorbidity network, which offers a holistic view of comorbidity patterns of CRC. A DG-guided self-attention mechanism is proposed to unearth novel dependencies beyond what DG offers, thus augmenting CRC survival prediction. Finally, each patient will be represented, and these representations will be used for survival prediction.Results: The experimental results show that SAGL outperforms state-of-the-art methods on a real-world dataset, with the receiver operating characteristic curve for 3- and 5-year survival prediction achieving 0.849±0.002 and 0.895±0.005, respectively. In addition, the comparison results with different graph neural network-based variants demonstrate the advantages of our DG-guided self-attention graph learning framework.Conclusions: Our study reveals that the potential of the DG-guided self-attention in optimizing feature graph learning which can improve the performance of CRC survival prediction.

KETERGANTUNGAN PETANI SAYURAN TERHADAP TENGKULAK SEBAGAI PATRON-KLIEN DALAM KEGIATAN PERTANIAN (STUDI KASUS DESA WAKULI KECAMATAN KAPONTORI KABUPATEN BUTON)

The income of vegetable farmers in Wakuli Village comes from production results and the prices received by farmers. Efficient management and increased production should have an impact on farmer welfare. However, in reality the exchange rate and the existence of farmers are often neglected in the economic system, especially in the distribution system for vegetable crops, so they are unable to provide positive profits. The aim of this research is to analyze the role of middlemen from the perspective of vegetable farmers and to determine the type of dependency of vegetable farmers as patron-clients. The informants in this research consisted of 1 key informant from village officials, the main informant consisted of 7 vegetable farmers and 3 middlemen, as well as 1 supporting informant from the head of the farmer group. The data collection techniques used in this research are observation, interviews and documentation. The instruments in this research were an interview guide, writing tools and recording equipment. Data analysis uses the Miles and Huberman interactive data analysis model. The research results show that middlemen in agricultural activities in Wakuli Village have an active role as capital owners, buyers and marketers. There are several types of middleman roles that are considered to be able to help farmers in dealing with economic and social problems. The role of the middleman covers all aspects (production, marketing, financial and social) needed by farmers in Wakuli Village. In the vegetable production process, farmers depend on capital provided by middlemen and in the marketing process the vegetables are sold to middlemen. Of all the farmers who were informants in this research, on average, they channeled their vegetable harvests to middlemen.

Smartphone dependency latent profile classification and association with emotional and behavioral difficulties among high school students in Korea

Purpose: The pervasive integration of smartphones into adolescents’ daily lives has resulted in a concerning upsurge in smartphone dependency among high school students. Due to the diverse types and severity levels of smartphone dependency, there is a critical need for research to explore its heterogeneity. This study aimed to identify latent profile of smartphone dependency in high school students based on the four subdomains of smartphone dependency: disturbance of adaptive functions, virtual life orientation, withdrawal, and tolerance. In addition, we explored how emotional and behavioral difficulties differ according to the profiles. Methods: We used data from 2,195 Korean high school students from the Korean Children and Youth Panel Survey 2018. Latent profile analysis (LPA) was performed to identify smartphone dependency latent profile. Statistical analysis including chi-square test, Analysis of Variance (ANOVA), and ranked Analysis of Covariance (ANCOVA) confirmed differences in smartphone use characteristics and emotional-behavioral difficulties according to the classified latent profile. Results: LPA identified four distinct latent profiles of smartphone dependency among high school students; 1) Underdependent type, 2) Moderate type, 3) Habitual user type, and 4) Virtual space dependent type. The results of ranked ANCOVA, controlling for gender, geographical location, economic status, parental smartphone dependency, and relational variables, revealed that habitual user type exhibited significantly higher rates of attention deficit hyperactivity disorder, social withdrawal, and depressive symptoms compared to other types. Conclusion: The identification of these profiles provides a foundation for developing tailored intervention programs for adolescents with different levels and patterns of smartphone dependency.

Minimal coverage of generalized typed inclusion dependencies in databases

The paper discusses the theory and algorithms necessary to construct a minimal covering of generalized typed inclusion dependencies. Traditionally, the construction of minimal covering is used for all types of dependencies in order to obtain a non-redundant and consistent database design. Generalized inclusion dependencies correspond to referential integrity constraints, when several main and several external relations are involved in one constraint, which corresponds to an ultragraph edge. In previous work, based on a study of the properties of dependencies, a system of axioms was presented with proof of consistency and completeness. In this work, the closures were studied for generalized typed inclusion dependencies. An algorithm for constructing closures has been developed and its correctness has been proven. The results obtained are further used to develop an algorithm for constructing a minimal covering. At the end of the article, examples are presented that demonstrate the operation of the algorithms.

Simulation-based sample size calculations of marginal proportional means models for recurrent events with competing risks.

In randomised controlled trials, the outcome of interest could be recurrent events, such as hospitalisations for heart failure. If mortality rates are non-negligible, both recurrent events and competing terminal events need to be addressed when formulating the estimand and statistical analysis is no longer trivial. In order to design future trials with primary recurrent event endpoints with competing risks, it is necessary to be able to perform power calculations to determine sample sizes. This paper introduces a simulation-based approach for power estimation based on a proportional means model for recurrent events and a proportional hazards model for terminal events. The simulation procedure is presented along with a discussion of what the user needs to specify to use the approach. The method is flexible and based on marginal quantities which are easy to specify. However, the method introduces a lack of a certain type of dependence. This is explored in a sensitivity analysis which suggests that the power is robust in spite of that. Data from a randomised controlled trial, LEADER, is used as the basis for generating data for a future trial. Finally, potential power gains of recurrent event methods as opposed to first event methods are discussed.

ДЖОНШЕРОВСКИЙ УНИВЕРСАЛЬНЫЙ ДИНАМИЧЕСКИЙ ОТКЛИК В ТОНКИХ ПЛЕНКАХ ПОЛИДИФЕНИЛЕНФТАЛИДА

It is known that the supramolecular structure of organic materials can have a significant effect on the conductivity of polymer films in both constant and alternating fields. In particular, the presence of associative supramolecular formations can lead to a significant increase in the local dielectric constant. At the same time, the presence of different types of supramolecular ordering can lead to the predominance of certain charge transfer mechanisms in polymer films, and, consequently, to the presence of a dependence of conductivity on the electric field strength. The degree of molecular ordering, both on the surface and in the bulk of a polymer film, can also have a significant effect on polarization. In this regard, studying the influence of the supramolecular structure on the conductivity of polymer films in an alternating electric field is an urgent task.Previously, using various methods, it was established that the supramolecular structure of films with a thickness of more than 300 nm is determined mainly by supramolecular formations of an associative structure. The structure of films with a thickness of 100 to 300 nm is determined by the mixed structure of globular and associative formations. The structure of films less than 100 nm thick is determined by supramolecular formations of a globular. It was also established by the method of thermally stimulated depolarization that three regions are observed on the dependence of the activation energy of capture centers on the thickness of the polymer film, corre- sponding to different supramolecular organizations.In this work, the dependence of the conductivity of poly (diphenylene phthalide) films in an alternating electric field in the low-frequency range from 0.1 to 106 Hz on the supramolecular structure of polymer films was studied. It was found that the type of dependence of the conductivity of polymer films on the frequency of an alternating field depends on the thickness and supramolecular structure of the polymer films, as well as the electric field strength. The applicability of the Jonsher’s model, which describes the conductivity of amorphous materials in an alternating electric field, for thin polymer films is discussed. A correlation between the power-law dependence index and the supramolecular structure of the polymer film and the influence of dipole polarization and injection of charge carriers into the bulk of the polymer film on the conductivity of the polymer film have been revealed.

Staged trees and asymmetry-labeled DAGs

AbstractBayesian networks are a widely-used class of probabilistic graphical models capable of representing symmetric conditional independence between variables of interest using the topology of the underlying graph. For categorical variables, they can be seen as a special case of the much more general class of models called staged trees, which can represent any non-symmetric conditional independence. Here we formalize the relationship between these two models and introduce a minimal Bayesian network representation of a staged tree, which can be used to read conditional independences intuitively. A new labeled graph termed asymmetry-labeled directed acyclic graph is defined, with edges labeled to denote the type of dependence between any two random variables. We also present a novel algorithm to learn staged trees which only enforces a specific subset of non-symmetric independences. Various datasets illustrate the methodology, highlighting the need to construct models that more flexibly encode and represent non-symmetric structures.

Аргументативное развертывание текста парламентских выступлений на микроуровне (на материале французского языка)

The research aims to identify functional types of semantic dependence within a fragment of the text of parliamentary speeches – the argumentative component. The paper examines the notion of an argumentative component and identifies its main elements. To establish an intentionally determined semantic dependence within this text segment, the notion of an argumentative move (AM) is introduced, which is understood as a functional unit of the addresser’s general communicative strategy and which is aimed at realizing a specific communicative intention. The relations of the functional-semantic connection, updating its argumentative moves, as well as the linguistic means that represent them are described. The scientific novelty of the work lies in the fact that based on the material of the French language, it typologizes the relations of semantic dependence at the micro level of the argumentative construction of the text of parliamentary speeches, taking into account the functional-communicative aspect. As a result of the research, the intentionally determined semantic relations typical for the corpus of texts under study were determined: proper causation (AM ‘causal basis’), nonproper causation (AM ‘argument’), consequence (AM ‘conclusion’), concession (AM ‘insufficient basis’), disjunctiveness (AM ‘expanded alternative’ and AM ‘alternative motivation’), their linguistic markers were identified and systematized.

Features of the photophoretic motion of an evaporating droplet in a viscous non-isothermal binary gas medium

A theoretical description of the photophoretic motion in a viscous nonisothermal binary gas mixture of a large evaporating spherical droplet with significant relative temperature differences in its vicinity is carried out in the quasi-stationary approximation for small Reynolds and Pecle numbers. When describing the properties of a gaseous medium, a power-law type of dependence of the coefficients of molecular transport (viscosity, diffusion and thermal conductivity) and density on temperature was taken into account. Numerical estimates have shown the nonlinear nature of the dependence of the photophoretic force and velocity on the average temperature of the droplet surface.

Global domain adaptation attention with data-dependent regulator for scene segmentation.

Most semantic segmentation works have obtained accurate segmentation results through exploring the contextual dependencies. However, there are several major limitations that need further investigation. For example, most approaches rarely distinguish different types of contextual dependencies, which may pollute the scene understanding. Moreover, local convolutions are commonly used in deep learning models to learn attention and capture local patterns in the data. These convolutions operate on a small neighborhood of the input, focusing on nearby information and disregarding global structural patterns. To address these concerns, we propose a Global Domain Adaptation Attention with Data-Dependent Regulator (GDAAR) method to explore the contextual dependencies. Specifically, to effectively capture both the global distribution information and local appearance details, we suggest using a stacked relation approach. This involves incorporating the feature node itself and its pairwise affinities with all other feature nodes within the network, arranged in raster scan order. By doing so, we can learn a global domain adaptation attention mechanism. Meanwhile, to improve the features similarity belonging to the same segment region while keeping the discriminative power of features belonging to different segments, we design a data-dependent regulator to adjust the global domain adaptation attention on the feature map during inference. Extensive ablation studies demonstrate that our GDAAR better captures the global distribution information for the contextual dependencies and achieves the state-of-the-art performance on several popular benchmarks.