Transparent Reasoning Research Articles

Triact package for R: analyzing the lying behavior of cows from accelerometer data.

Accelerometers are sensors proven to be useful to analyze the lying behavior of cows. For reasons of algorithm transparency and control, researchers often prefer to use their own data analysis scripts rather than proprietary software. We developed the triact R package that assists animal scientists in analyzing the lying behavior of cows from raw data recorded with a triaxial accelerometer (manufacturer agnostic) attached to a hind leg. In a user-friendly workflow, triact allows the determination of common measures for lying behavior including total lying duration, the number of lying bouts, and the mean duration of lying bouts. Further capabilities are the description of lying laterality and the calculation of proxies for the level of physical activity of the cow. In this publication we describe the functionality of triact and the rationales behind the implemented algorithms. The triact R package is developed as an open-source project and freely available via the CRAN repository.

ARDST: An Adversarial‐Resilient Deep Symbolic Tree for Adversarial Learning

The advancement of intelligent systems, particularly in domains such as natural language processing and autonomous driving, has been primarily driven by deep neural networks (DNNs). However, these systems exhibit vulnerability to adversarial attacks that can be both subtle and imperceptible to humans, resulting in arbitrary and erroneous decisions. This susceptibility arises from the hierarchical layer‐by‐layer learning structure of DNNs, where small distortions can be exponentially amplified. While several defense methods have been proposed, they often necessitate prior knowledge of adversarial attacks to design specific defense strategies. This requirement is often unfeasible in real‐world attack scenarios. In this paper, we introduce a novel learning model, termed “immune” learning, known as adversarial‐resilient deep symbolic tree (ARDST), from a neurosymbolic perspective. The ARDST model is semiparametric and takes the form of a tree, with logic operators serving as nodes and learned parameters as weights of edges. This model provides a transparent reasoning path for decision‐making, offering fine granularity, and has the capacity to withstand various types of adversarial attacks, all while maintaining a significantly smaller parameter space compared to DNNs. Our extensive experiments, conducted on three benchmark datasets, reveal that ARDST exhibits a representation learning capability similar to DNNs in perceptual tasks and demonstrates resilience against state‐of‐the‐art adversarial attacks.

Optimized Self-Similar Borel Summation

The method of Fractional Borel Summation is suggested in conjunction with self-similar factor approximants. The method used for extrapolating asymptotic expansions at small variables to large variables, including the variables tending to infinity, is described. The method is based on the combination of optimized perturbation theory, self-similar approximation theory, and Borel-type transformations. General Borel Fractional transformation of the original series is employed. The transformed series is resummed in order to adhere to the asymptotic power laws. The starting point is the formulation of dynamics in the approximations space by employing the notion of self-similarity. The flow in the approximation space is controlled, and “deep” control is incorporated into the definitions of the self-similar approximants. The class of self-similar approximations, satisfying, by design, the power law behavior, such as the use of self-similar factor approximants, is chosen for the reasons of transparency, explicitness, and convenience. A detailed comparison of different methods is performed on a rather large set of examples, employing self-similar factor approximants, self-similar iterated root approximants, as well as the approximation technique of self-similarly modified Padé–Borel approximations.

Concise rule induction algorithm based on one-sided maximum decision tree approach

As the importance of machine learning tools for decision support continues to grow, interpretability has emerged as a key factor. Rule-based classification algorithms, such as decision trees and rule induction, enable high local interpretability by providing transparent reasoning rules in an IF-THEN format. In this context, it is essential to provide concise and clear rules and conditions to achieve high local interpretability. This study proposes a novel Concise Algorithm, designed to effectively remove irrelevant conditions from classification rules. We present a framework incorporating the Concise Algorithm, which employs the One-Sided-Maximum decision tree algorithm for rule generation, followed by the application of the Concise Algorithm to remove irrelevant conditions. This proposed framework produces a rule-based classification model that exhibits an enhanced predictive performance-interpretability trade-off compared to benchmark methods (CART, Ripper, CN2, and modified One-Sided-Maximum), as demonstrated by empirical tests conducted on 19 UCI datasets. A case study focusing on the breast-cancer-wisconsin dataset provides a comprehensive analysis of the rule and condition generation processes.

Modern Artificial Intelligence-An Overview

Modern Artificial Intelligence (AI) is a rapidly evolving field that encompasses a range of techniques and approaches, including machine learning, deep learning, natural language processing, computer vision, robotics, and more. The development of AI technologies has enabled unprecedented levels of accuracy in tasks such as image and speech recognition, natural language understanding, and game playing. This has been made possible by the rise of deep learning, which involves training artificial neural networks on vast amounts of data to recognize patterns and make predictions with high accuracy. Other recent advances in modern AI include the development of generative models and reinforcement learning. Despite the significant progress made in modern AI, there are still many challenges that need to be addressed, including issues related to data privacy, fairness, and bias, and the need for more explainable AI systems that can provide clear and transparent reasoning for their decisions. This study provides an overview of modern AI and its applications, as well as the challenges and opportunities that lie ahead in this rapidly evolving field.

Engineering complexity beyond the surface: discerning the viewpoints, the drivers, and the challenges

Complexity is often regarded as a “problem” to solve. Instead of attempting to solve complexity, we follow systems engineering practices and switch back to the problem domain, where a major obstacle is the impossibility to universally define complexity. As a workaround, we explored complexity characterization and its existing shortcomings, including: lack of standardization, inconsistent semantics, system-centricity, insufficiently transparent reasoning, and lack of validation. To address these shortcomings, we proposed a compilatory framework to characterize complexity using the Five Ws information-gathering method. The answer to the WHO question proposed four complexity viewpoints; the answer to the WHY question proposed a two-dimensional structure for complexity drivers; and the answer to the WHAT question derived generalized complexity challenges. As a preliminary step to show the potential of the framework to characterize complexity, we used and validated it as a tool to structure general literature related to complexity. In general, our findings suggest that papers with complexity solutions do not frame their research within the complexity problem domain, hindering the contribution evaluation. Through the viewpoints, we identified general research gaps of six solution directions. From the drivers, we noted three observations in the discourse of complexity origins: (1) a system-driven tendency, (2) a preference for concreteness vs. abstraction, and (3) an unclear distinction between origins and effects. Through the challenges’ findings we explored two hypotheses: (1) a system-centric preference; and (2) a solution-oriented vision, both of which were supported by the results (most challenges relate to the system viewpoint and challenges are defined based on solution directions).

Can narratives about sovereign debt be generally ideologically suspicious? An exercise in broadening the scope of ideology critique

Can narratives about sovereign debt be generally ideologically suspicious? An exercise in broadening the scope of ideology critique

A Novel Belief Rule-Based Fault Diagnosis Method with Interpretability

Fault diagnosis plays an irreplaceable role in the normal operation of equipment. A fault diagnosis model is often required to be interpretable for increasing the trust between humans and the model. Due to the understandable knowledge expression and transparent reasoning process, the belief rule base (BRB) has extensive applications as an interpretable expert system in fault diagnosis. Optimization is an effective means to weaken the subjectivity of experts in BRB, where the interpretability of BRB may be weakened. Hence, to obtain a credible result, the weakening factors of interpretability in the BRB-based fault diagnosis model are firstly analyzed, which are manifested in deviation from the initial judgement of experts and over-optimization of parameters. For these two factors, three indexes are proposed, namely the consistency index of rules, consistency index of the rule base and over-optimization index, to measure the interpretability of the optimized model. Considering both the accuracy and interpretability of a model, an improved coordinate ascent (I-CA) algorithm is proposed to fine-tune the parameters of the fault diagnosis model based on BRB. In I-CA, the algorithm combined with the advance and retreat method and the golden section method is employed to be one-dimensional search algorithm. Furthermore, the random optimization sequence and adaptive step size are proposed to improve the accuracy of the model. Finally, a case study of fault diagnosis in aerospace relays based on BRB is carried out to verify the effectiveness of the proposed method.

A Self-Interpretable Deep Learning Model for Seizure Prediction Using a Multi-Scale Prototypical Part Network.

The epileptic seizure prediction (ESP) method aims to timely forecast the occurrence of seizures, which is crucial to improving patients' quality of life. Many deep learning-based methods have been developed to tackle this issue and achieve significant progress in recent years. However, the "black-box" nature of deep learning models makes the clinician mistrust the prediction results, severely limiting its clinical application. For this purpose, in this study, we propose a self-interpretable deep learning model for patient-specific epileptic seizure prediction: Multi-Scale Prototypical Part Network (MSPPNet). This model attempts to measure the similarity between the inputs and prototypes (learned during training) as evidence to make final predictions, which could provide a transparent reasoning process and decision basis (e.g., significant prototypes for inputs and corresponding similarity score). Furthermore, we assign different sizes to the prototypes in latent space to capture the multi-scale features of EEG signals. To the best of our knowledge, this is the first study that develops a self-interpretable deep learning model for seizure prediction, other than the existing post hoc interpretation studies. Our proposed model is evaluated on two public epileptic EEG datasets (CHB-MIT: 16 patients with a total of 85 seizures, Kaggle: 5 dogs with a total of 42 seizures), with a sensitivity of 93.8% and a false prediction rate of 0.054/h in the CHB-MIT dataset and a sensitivity of 88.6% and a false prediction rate of 0.146/h in the Kaggle dataset, achieving the current state-of-the-art performance with self-interpretable evidence.

Wie belastbar sind Studien der aktuell dauerhaft aufgenommenen digitalen Gesundheitsanwendungen (DiGA)? Methodische Qualität der Studien zum Nachweis positiver Versorgungseffekte von DiGA

IntroductionSince September 2020 digital health applications (DiGA) can be prescribed by physicians and psychotherapists and are reimbursed within the German Statutory Health Insurance (SHI) system for the first time worldwide. For full reimbursement, the manufacturers have to provide evidence based on scientific studies that the DiGA can provide ‘positive health care effects’. This study aims to analyze and evaluate the methodological quality of efficacy studies of DiGA in the categories ‘Nervensystem’ and ‘Psyche’ of the DiGA register that are permanently accepted. MethodsThe methodological quality was assessed using the revised Cochrane risk-of-bias tool for randomized trials (RoB 2). The risk of bias was assessed for the primary endpoint of each study according to an intention-to-treat analysis. ResultsSix DiGA were assessed for their methodological quality. Randomized controlled trials were conducted for all 6 DiGA that showed a high risk of bias, which was, in particular, due to a lack of blinding of the studies. In addition, drop-outs were significantly higher in the intervention group than in the control group in most studies. For most of the DiGA no published study protocol was available in advance so an analysis of a potential selective choice of the evaluation methodology was not possible. DiscussionFor reasons of transparency, verifiability, and comprehensibility of the study results, registration in a study registry and, more importantly, the publication of study protocols should be mandatory before the start of the studies. In addition, studies should be blinded by comparing the DiGA with a ‘sham application’ to reduce the high risk of bias. Differences in the drop-out rates of the investigated studies could indicate a lack of efficacy of the treatment in the intervention group, (technical) problems in the application of the DiGA, or a lack of motivation of the participants. ConclusionThe interim results 18 months after the introduction of DiGA in the German SHI system show that the studies on the evidence of the benefits of DiGA have a high potential for bias in certain areas. However, it should be positively emphasized that the manufacturers submitted randomized controlled trials to prove the medical benefit of the DiGAs investigated.

Упровадження концепції податкових витрат в Україні: проблеми та шляхи їх вирішення

Introduction. The preparation of reports on tax expenditures and their use in the process of budget management has become widespread in many countries of the world. Since 2004, the Global Tax Expenditure Database has included information on tax expenditures in Ukraine. However, the degree of implementation of the concept of tax expenditures in the budget process varies greatly in different countries. Problem Statement. Imperfect procedures for reporting on tax expenditures and their formal connection with the budget process, or even the absence of such a connection at all are significant reasons for insufficient transparency of tax expenditures budgets and the effectiveness of their use as public policy tools. Purpose. The goal is to assess the state, determine the problems of integrating the report on budget losses from the use of tax reliefs (tax expenditures) into the budget process in Ukraine and ways to solve them. Methods. General scientific methods of cognition were used: critical analysis of the controversial provisions of the concept of tax expenditures; a combination of abstract-logical and concrete, comparative analysis. Results. The steps that should be taken to ensure the effective implementation of the tax expenditure budget, as well as the state of their implementation in several countries and in Ukraine, are analyzed. The problems on the way to the implementation of the concept of tax expenditures in Ukraine are revealed: uncertainty at the legislative level of the concept of “tax expenditures” and the body responsible for the formation of a consolidated report on tax expenditures; non-disclosure of their assessment methodology; improper coordination of the work of the State Tax Service of Ukraine and the State Customs Service of Ukraine on collecting and transmitting information on tax expenditures; formal inclusion of information on tax expenditures in the budget process; absence of systematic evaluation of the effectiveness and efficiency of tax expenditures. Conclusions. It has been established that only the first steps have been taken in Ukraine towards the development and implementation of the tax expenditure budget and its integration into the country's budget process, and proposals have been developed for measures that must be implemented for further progress on this path, in particular changes to tax and budget legislation, as well as the administrative documents of executive authorities involved in the preparation of reports on tax expenditures.

A dominant set-informed interpretable fuzzy system for automated diagnosis of dementia.

Dementia is an incurable neurodegenerative disease primarily affecting the older population, for which the World Health Organisation has set to promoting early diagnosis and timely management as one of the primary goals for dementia care. While a range of popular machine learning algorithms and their variants have been applied for dementia diagnosis, fuzzy systems, which have been known effective in dealing with uncertainty and offer to explicitly reason how a diagnosis can be inferred, sporadically appear in recent literature. Given the advantages of a fuzzy rule-based model, which could potentially result in a clinical decision support system that offers understandable rules and a transparent inference process to support dementia diagnosis, this paper proposes a novel fuzzy inference system by adapting the concept of dominant sets that arise from the study of graph theory. A peeling-off strategy is used to iteratively extract from the constructed edge-weighted graph a collection of dominant sets. Each dominant set is further converted into a parameterized fuzzy rule, which is finally optimized in a supervised adaptive network-based fuzzy inference framework. An illustrative example is provided that demonstrates the interpretable rules and the transparent reasoning process of reaching a decision. Further systematic experiments conducted on data from the Open Access Series of Imaging Studies (OASIS) repository, also validate its superior performance over alternative methods.

Validation of the hen's egg test for micronucleus induction (HET-MN): detailed protocol including scoring atlas, historical control data and statistical analysis.

A validation exercise of the hen’s egg test for micronucleus induction was finalised with a very good predictivity based on the analysis of micronuclei in peripheral erythrocytes of fertilised chicken eggs (Reisinger et al. The hen’s egg test for micronucleus-induction (HET-MN): validation data set. Mutagenesis, this issue). For transparency reasons this complementary publication provides further details on the assay especially as it was the first validation study in the field of genotoxicity testing involving the use of chicken eggs. Thus, the experimental protocol is described in detail and is complemented by a scoring atlas for microscopic analysis in blood cells. In addition, general characteristics of the test system, which is able to mirror the systemic availability of test compounds, are delineated: the test compound passes the egg membrane and is taken up by the blood vessels of the underlying chorioallantoic membrane. Subsequently, it is distributed by the circulating blood, metabolised by the developing liver and the yolk sac membrane and finally excreted into the allantois, a bladder equivalent. In specific, the suitability of the test system for genotoxicity testing is shown by, inter alia, a low background DNA damage in a comprehensive historical control database. In addition, the state-of-the-art statistical method used to evaluate obtained data is delineated. It combines laboratory-specific effect threshold with the Umbrella–Williams test, a statistical model also of interest for other genotoxicity test methods.

Pursuing open-source development of predictive algorithms: the case of criminal sentencing algorithms

Currently, there is uncertainty surrounding the merits of open-source versus proprietary algorithm development. Though justification in favor of each exists, we argue that open-source algorithm development should be the standard in highly consequential contexts that affect people’s lives for reasons of transparency and collaboration, which contribute to greater predictive accuracy and enjoy the additional advantage of cost-effectiveness. To make this case, we focus on criminal sentencing algorithms, as criminal sentencing is highly consequential, and impacts society and individual people. Furthermore, the popularity of this topic has surged in the wake of recent studies uncovering racial bias in proprietary sentencing algorithms among other issues of over-fitting and model complexity. We suggest that these issues are exacerbated by the proprietary and expensive nature of virtually all widely used criminal sentencing algorithms. Upon replicating a major algorithm using real criminal profiles, we fit three penalized regressions and demonstrate an increase in predictive power of these open-source and relatively computationally inexpensive options. The result is a data-driven suggestion that if judges who are making sentencing decisions want to craft appropriate sentences based on a high degree of accuracy and at low costs, then they should be pursuing open-source options.

Are there systematic barriers to participation in cancer treatment trials by Aboriginal and Torres Strait Islander cancer patients in Australia?

ObjectiveTo identify factors that may systematically reduce opportunities for Aboriginal and Torres Strait Islander Australians to participate in cancer clinical trials. MethodsAnalysis of online documents from the Australia and New Zealand Clinical Trials Registry for cancer treatment trials (Phase 3, 4 or Not Applicable) with at least one Australian site, registered in 2014–2018. ResultsAmong 365 eligible trials, most (89%) had sites only in major cities/inner regional areas, but 39% of Aboriginal and Torres Strait Islander Australians live outside these areas. Seven cancer types accounted for 58% of cancers among Aboriginal and Torres Strait Islander people, but only 46% of trials addressed these cancers. Most trials specified exclusions relating to comorbidities/health status. A substantial minority of trials (38%) explicitly referred to investigator opinion/judgment as a relevant determinant of patient eligibility. ConclusionAboriginal and Torres Strait Islander patients appear to have a reduced opportunity to participate in trials because of where they live, their type of cancer and their general health status, as well as for less transparent reasons relating to investigator judgment. Implications for public healthGreater transparency and greater scrutiny of barriers to trial participation for Aboriginal and Torres Strait Islander Australians are needed to ensure equitable access.

Enhancement of the European Parliament's Monitoring for Better Coherence between Trade Policy and Non-Trade Policy Objectives (Ntpos)

Since the Treaty of Lisbon, trade policy has become an explicit part of the EU's external policy and integrated into the general framework of the EU´s external policy, but must also be in conformity with internal policies. Thus, trade policy is subject to a requirement of multiple coherence. Beyond constitutional obligations, other drivers work for the inclusion of non-genuine commercial policy objectives in trade policy, such as the orientation of contemporary trade politics towards the behind the border issues of national regulation, so that trade policy became closely intertwined with domestic regulatory policy. Therefore the actors primarily responsible for legislation, i.e. parliaments, advocate for their extended participation in determining trade policy, and rightly so for reasons of transparency, control and political inclusiveness. Parliaments thus become actors of respect for and positive consideration of non-commercial policy objectives in trade policy, which applies as well to the European Parliament (EP). Hence, an institutional design of policy formulation cycles and decision-making in EU trade policy that strives for better coherence of trade concerns with NTPOs must focus on strengthening the influence of the EP and improve its participatory rights in decision-making and its control and monitoring mechanisms. Consequently, the present paper derives proposals for improving EP´s monitoring mechanisms for the benefit of non-trade policy objectives (NTPOs) in trade policy from an analysis of weaknesses in the negotiation and implementation stage of trade policy.

Conflict resolution in post-violence societies: some guidance for the judiciary

ABSTRACT The article is concerned with a worldwide phenomenon that is particularly prevalent in post-violence societies, whereby conflicts that should have been resolved in the political arena, are adjudicated by the judiciary. It offers reasons for this phenomenon and argues that when judges face this task, they tend to base their reasoning on human rights arguments. While convenient, using human rights as proxies for a more complex political analysis can undermine the quality of both the judgment and remedies provided by the court. Alternatively, when courts adjudicate political conflicts, they should rely on constitutional guiding principles. Their main advantage is that they expressly acknowledge the political and controversial nature of the conflicts at hand, thus result in a more transparent reasoning. Additionally, their more flexible nature makes constitutional principles applicable to a broader range of conflicts and can result in more appropriate remedies upon their adjudication.

De Los idiotas a La gallina degollada: el encuentro con el horror

In this article, we aim to carry out a comparative analysis of the short stories The Idiots, by Joseph Conrad, and La gallina degollada, by Horacio Quiroga, whereby we introduce the term conjunction; in this case, the conjunction of horror. To this end, we refer to the key actions that are common to the two stories. For reasons of transparency, we avoid the excess of theories regarding intertextual relations, although we allude to one key theoretical concept for this study: the parody, proposed by Bakhtin. Our purpose, however, is to emphasize our own conceptual proposal: that of the conjunction. It is our intention that the readers face themselves with the antecedent story which gives birth to the succeeding story, and reflect on the creative process, on the poetics itself. As the axis of articulation of this article, we highlight the influence of Conrad on Quiroga, and how both writers, in the above-mentioned stories, conceive the theme of horror based on the motif of the brothers who suffer physical and mental degradation, which trigger tragedies for the young couples.

Trust in Automation as a Function of Transparency and Teaming

Information analysis is a critical piece of successful operations. As automation becomes a more common element in this analysis, it is important to understand what information should be communicated to users to promote proper use of automation. Both transparency, information regarding how and why automation functions proceed, and uncertainty may be two vital pieces of communication. Method: Twenty-six participants completed a simulated tank identification task with an automated or human teammate across 3 levels of teammate transparency. Trials varied in presence or absence of uncertainty information. Trust and performance effects were evaluated. Users were shown a terrain image and had to determine if a military tank was present in the terrain. The teammate provided a Tank Present or Absent recommendation along with transparent reasoning for the recommendation. Participants then made a tank presence judgement. Results: Transparency increased user trust as well as reaction times. Having uncertainty information increased user trust, but decreased performance accuracy. Teammate type had no significant effect on trust, accuracy, or reaction time. Discussion: The data indicate that transparency and uncertainty information can help users form an accurate mental model of teammate capabilities.

The increasing complexity of the core outcomes landscape

In this project, we set out to identify ways to increase the uptake of core outcome sets in clinical research. In doing so, we uncovered a growing recognition, across many different health care sectors, of the need for common, relevant outcomes to improve the quality of decision-making. This has led to a plethora of projects, initiatives, and new organizations all intended to develop standardized outcomes and outcome measures for their particular fields. However, the standardized outcome sets developed across siloed initiatives do not carry over to other sectors, such as from research to quality of care. This trend has the potential to lead to confusion and unintended redundancies, as well as wasteful use of both financial and intellectual resources. Better communication and collaboration among different initiatives, and more deliberate alignments of initiative scopes, are needed to ensure a future paradigm in which standards align across contexts where possible and differ for understandable and transparent reasons.