Search

The multivariate time series classification (MTSC) task aims to predict a class label for a given time series. Recently, modern deep learning-based approaches have achieved promising performance over traditional methods for MTSC tasks. The success of these approaches relies on access to the massive amount of labeled data (i.e., annotating or assigning tags to each sample that shows its corresponding category). However, obtaining a massive amount of labeled data is usually very time-consuming and expensive in many real-world applications such as medicine, because it requires domain experts’ knowledge to annotate data. Insufficient labeled data prevents these models from learning discriminative features, resulting in poor margins that reduce generalization performance. To address this challenge, we propose a novel approach: supervised contrastive learning for time series classification (SupCon-TSC). This approach improves the classification performance by learning the discriminative low-dimensional representations of multivariate time series, and its end-to-end structure allows for interpretable outcomes. It is based on supervised contrastive (SupCon) loss to learn the inherent structure of multivariate time series. First, two separate augmentation families, including strong and weak augmentation methods, are utilized to generate augmented data for the source and target networks, respectively. Second, we propose the instance-level, and cluster-level SupCon learning approaches to capture contextual information to learn the discriminative and universal representation for multivariate time series datasets. In the instance-level SupCon learning approach, for each given anchor instance that comes from the source network, the low-variance output encodings from the target network are sampled as positive and negative instances based on their labels. However, the cluster-level approach is performed between each instance and cluster centers among batches, as opposed to the instance-level approach. The cluster-level SupCon loss attempts to maximize the similarities between each instance and cluster centers among batches. We tested this novel approach on two small cardiopulmonary exercise testing (CPET) datasets and the real-world UEA Multivariate time series archive. The results of the SupCon-TSC model on CPET datasets indicate its capability to learn more discriminative features than existing approaches in situations where the size of the dataset is small. Moreover, the results on the UEA archive show that training a classifier on top of the universal representation features learned by our proposed method outperforms the state-of-the-art approaches.

Shared parking plays a crucial role in alleviating parking pressure, but the heterogeneity of potential suppliers’ intentions was often ignored. This study addresses this gap by adopting an interpretable Machine Learning (ML) framework to investigate parking space sharing intentions, considering individual differences. A survey with 383 respondents from mainland China was conducted, and a Latent Class Model (LCM) identified three distinct groups of potential suppliers. The Light Gradient Boosting Machine (LightGBM), outperforming other ML models, was used to quantify factors influencing sharing behaviors. The SHapley Additive exPlanation (SHAP) approach revealed that influential factors vary across different latent classes. These findings provide insights for shared parking operators to encourage potential suppliers’ participation in shared parking. [ABSTRACT FROM AUTHOR]

In recent years, the question of the reliability of Machine Learning (ML) methods has acquired significant importance, and the analysis of the associated uncertainties has motivated a growing amount of research. However, most of these studies have applied standard error analysis to ML models—and in particular Deep Neural Network (DNN) models—which represent a rather significant departure from standard scientific modelling. It is therefore necessary to integrate the standard error analysis with a deeper epistemological analysis of the possible differences between DNN models and standard scientific modelling and the possible implications of these differences in the assessment of reliability. This article offers several contributions. First, it emphasises the ubiquitous role of model assumptions (both in ML and traditional science) against the illusion of theory-free science. Secondly, model assumptions are analysed from the point of view of their (epistemic) complexity, which is shown to be language-independent. It is argued that the high epistemic complexity of DNN models hinders the estimate of their reliability and also their prospect of long term progress. Some potential ways forward are suggested. Thirdly, this article identifies the close relation between a model’s epistemic complexity and its interpretability, as introduced in the context of responsible AI. This clarifies in which sense—and to what extent—the lack of understanding of a model (black-box problem) impacts its interpretability in a way that is independent of individual skills. It also clarifies how interpretability is a precondition for a plausible assessment of the reliability of any model, which cannot be based on statistical analysis alone. This article focuses on the comparison between traditional scientific models and DNN models. However, Random Forest (RF) and Logistic Regression (LR) models are also briefly considered. [ABSTRACT FROM AUTHOR]

Mineral prospectivity mapping (MPM) is designed to reduce the exploration search space by combining and analyzing geological prospecting big data. Such geological big data are too large and complex for humans to effectively handle and interpret. Artificial intelligence (AI) algorithms, which are powerful tools for mining nonlinear mineralization patterns in big data obtained from mineral exploration, have demonstrated excellent performance in MPM. However, AI-driven MPM faces several challenges, including difficult interpretability, poor generalizability, and physical inconsistencies. In this study, based on previous studies, we devised a novel workflow that aims to constructing more transparent and explainable artificial intelligence (XAI) models for MPM by embedding domain knowledge throughout the AI-driven MPM, from input data to model design and model output. This newly proposed approach provides strong geological and conceptual leads that guide the entire AI-driven MPM model training process, thereby improving model interpretability and performance. Overall, the development of XAI models for MPM is capable of embedding prior and expert knowledge throughout the modeling process, presenting a valuable and promising area for future research designed to improve MPM. [ABSTRACT FROM AUTHOR]

Machine learning (ML) models are widely employed for crash severity modelling, yet their interpretability remains underexplored. Interpretation is crucial for comprehending ML results and aiding informed decision-making. This study aims to implement an interpretable ML to visualize the impacts of factors on crash severity using 5 years of freeways data from Iran. Methods including classification and regression trees (CART), K-nearest neighbours (KNNs), random forest (RF), artificial neural network (ANN) and support vector machines (SVM) were applied, with RF demonstrating superior accuracy, recall, F1-score and ROC. The accumulated local effects (ALE) were utilized for interpretation. Findings suggest that light traffic conditions ( volume / capacity < 0.5) with critical values around 0.05 or 0.38, and higher proportion of large trucks and buses, particularly at 10% and 4%, are associated with severe crashes. Additionally, speeds exceeding 90 km/h, drivers younger than 30 years, rollover crashes, collisions with fixed objects and barriers, nighttime driving and driver fatigue elevate the likelihood of severe crashes. [ABSTRACT FROM AUTHOR]

The interpretability of AI is just as important as its performance. In the LegalAI field, there have been efforts to enhance the interpretability of models, but a trade-off between interpretability and prediction accuracy remains inevitable. In this paper, we introduce a novel framework called LK-IB for compulsory measure prediction (CMP), one of the critical tasks in LegalAI. LK-IB leverages Legal Knowledge and combines an Interpretable model and a Black-box model to balance interpretability and prediction performance. Specifically, LK-IB involves three steps: (1) inputting cases into the first module, where first-order logic (FOL) rules are used to make predictions and output them directly if possible; (2) sending cases to the second module if FOL rules are not applicable, where a case distributor categorizes them as either "simple" or "complex"; and (3) sending simple cases to an interpretable model with strong interpretability and complex cases to a black-box model with outstanding performance. Experimental results demonstrate that the LK-IB framework provides more interpretable and accurate predictions than other state-of-the-art models. Given that the majority of cases in LegalAI are simple, the idea of model combination has significant potential for practical applications. [ABSTRACT FROM AUTHOR]

The increasing prevalence of omics data sources is pushing the study of regulatory mechanisms underlying complex diseases such as cancer. However, the vast quantities of molecular features produced and the inherent interplay between them lead to a level of complexity that hampers both descriptive and predictive tasks, requiring custom‐built algorithms that can extract relevant information from these sources of data. We propose a transformation that moves data centered on molecules (e.g., transcripts and proteins) to a new data space focused on putative regulatory modules given by statistically relevant co‐expression patterns. To this end, the proposed transformation extracts patterns from the data through biclustering and uses them to create new variables with guarantees of interpretability and discriminative power. The transformation is shown to achieve dimensionality reductions of up to 99% and increase predictive performance of various classifiers across multiple omics layers. Results suggest that omics data transformations from gene‐centric to pattern‐centric data supports both prediction tasks and human interpretation, notably contributing to precision medicine applications. [ABSTRACT FROM AUTHOR]

In this paper, our goal is to enhance the interpretability of Generalized Linear Models by identifying the most relevant interactions between categorical predictors. Searching for interaction effects can quickly become a highly combinatorial, and thus computationally costly, problem when we have many categorical predictors or even a few of them but with many categories. Moreover, the estimation of coefficients requires large training samples with enough observations for each interaction between categories. To address these bottlenecks, we propose to find a reduced representation for each categorical predictor as a binary predictor, where categories are clustered based on a dissimilarity. We provide a collection of binarized representations for each categorical predictor, where the dissimilarity takes into account information from the main effects and the interactions. The choice of the binarized predictors representing the categorical predictors is made with a novel heuristic procedure that is guided by the accuracy of the so-called binarized model. We test our methodology on both real-world and simulated data, illustrating that, without damaging the out-of-sample accuracy, our approach trains sparse models including only the most relevant interactions between categorical predictors. [ABSTRACT FROM AUTHOR]

Linear least squares is one of the most widely used regression methods in many fields. The simplicity of the model allows this method to be used when data is scarce and allows practitioners to gather some insight into the problem by inspecting the values of the learnt parameters. In this paper we propose a variant of the linear least squares model allowing practitioners to partition the input features into groups of variables that they require to contribute similarly to the final result. We show that the new formulation is not convex and provide two alternative methods to deal with the problem: one non-exact method based on an alternating least squares approach; and one exact method based on a reformulation of the problem. We show the correctness of the exact method and compare the two solutions showing that the exact solution provides better results in a fraction of the time required by the alternating least squares solution (when the number of partitions is small). We also provide a branch and bound algorithm that can be used in place of the exact method when the number of partitions is too large as well as a proof of NP-completeness of the optimization problem. [ABSTRACT FROM AUTHOR]

With the rise of artificial intelligence in our everyday lives, the need for human interpretation of machine learning models' predictions emerges as a critical issue. Generally, interpretability is viewed as a binary notion with a performance trade-off. Either a model is fully-interpretable but lacks the ability to capture more complex patterns in the data, or it is a black box. In this paper, we argue that this view is severely limiting and that instead interpretability should be viewed as a continuous domain-informed concept. We leverage the well-known Mixture of Experts architecture with user-defined limits on non-interpretability. We extend this idea with a counterfactual fairness module to ensure the selection of consistently fair experts: FairMOE. We perform an extensive experimental evaluation with fairness-related data sets and compare our proposal against state-of-the-art methods. Our results demonstrate that FairMOE is competitive with the leading fairness-aware algorithms in both fairness and predictive measures while providing more consistent performance, competitive scalability, and, most importantly, greater interpretability. [ABSTRACT FROM AUTHOR]

Graph Neural Networks (GNNs) are a popular class of machine learning models. Inspired by the learning to explain (L2X) paradigm, we propose L2xGnn, a framework for explainable GNNs which provides faithful explanations by design. L2xGnn learns a mechanism for selecting explanatory subgraphs (motifs) which are exclusively used in the GNNs message-passing operations. L2xGnn is able to select, for each input graph, a subgraph with specific properties such as being sparse and connected. Imposing such constraints on the motifs often leads to more interpretable and effective explanations. Experiments on several datasets suggest that L2xGnn achieves the same classification accuracy as baseline methods using the entire input graph while ensuring that only the provided explanations are used to make predictions. Moreover, we show that L2xGnn is able to identify motifs responsible for the graph's properties it is intended to predict. [ABSTRACT FROM AUTHOR]

State-of-the-art deep neural networks (DNNs) are highly effective at tackling many real-world tasks. However, their widespread adoption in mission-critical contexts is limited due to two major weaknesses - their susceptibility to adversarial attacks and their opaqueness. The former raises concerns about DNNs' security and generalization in real-world conditions, while the latter, opaqueness, directly impacts interpretability. The lack of interpretability diminishes user trust as it is challenging to have confidence in a model's decision when its reasoning is not aligned with human perspectives. In this research, we (1) examine the effect of adversarial robustness on interpretability, and (2) present a novel approach for improving DNNs' interpretability that is based on the regularization of neural activation sensitivity. We evaluate the interpretability of models trained using our method to that of standard models and models trained using state-of-the-art adversarial robustness techniques. Our results show that adversarially robust models are superior to standard models, and that models trained using our proposed method are even better than adversarially robust models in terms of interpretability.(Code provided in supplementary material.) [ABSTRACT FROM AUTHOR]

Feature attribution maps are a popular approach to highlight the most important pixels in an image for a given prediction of a model. Despite a recent growth in popularity and available methods, the objective evaluation of such attribution maps remains an open problem. Building on previous work in this domain, we investigate existing quality metrics and propose new variants of metrics for the evaluation of attribution maps. We confirm a recent finding that different quality metrics seem to measure different underlying properties of attribution maps, and extend this finding to a larger selection of attribution methods, quality metrics, and datasets. We also find that metric results on one dataset do not necessarily generalize to other datasets, and methods with desirable theoretical properties do not necessarily outperform computationally cheaper alternatives in practice. Based on these findings, we propose a general benchmarking approach to help guide the selection of attribution methods for a given use case. Implementations of attribution metrics and our experiments are available online (https://github.com/arnegevaert/benchmark-general-imaging). [ABSTRACT FROM AUTHOR]

Interpretable representations are the backbone of many explainers that target black-box predictive systems based on artificial intelligence and machine learning algorithms. They translate the low-level data representation necessary for good predictive performance into high-level human-intelligible concepts used to convey the explanatory insights. Notably, the explanation type and its cognitive complexity are directly controlled by the interpretable representation, tweaking which allows to target a particular audience and use case. However, many explainers built upon interpretable representations overlook their merit and fall back on default solutions that often carry implicit assumptions, thereby degrading the explanatory power and reliability of such techniques. To address this problem, we study properties of interpretable representations that encode presence and absence of human-comprehensible concepts. We demonstrate how they are operationalised for tabular, image and text data; discuss their assumptions, strengths and weaknesses; identify their core building blocks; and scrutinise their configuration and parameterisation. In particular, this in-depth analysis allows us to pinpoint their explanatory properties, desiderata and scope for (malicious) manipulation in the context of tabular data where a linear model is used to quantify the influence of interpretable concepts on a black-box prediction. Our findings lead to a range of recommendations for designing trustworthy interpretable representations; specifically, the benefits of class-aware (supervised) discretisation of tabular data, e.g., with decision trees, and sensitivity of image interpretable representations to segmentation granularity and occlusion colour. [ABSTRACT FROM AUTHOR]

Mining counterfactual antecedents became a valuable tool to discover knowledge and explain machine learning models. It consists of generating synthetic samples from an original sample to achieve the desired outcome in a machine learning model, thus helping to understand the prediction. An insightful methodology would explore a broader set of counterfactual antecedents to reveal multiple possibilities while operating on any classifier. Therefore, we create a tree-based search that requires monotonicity from the objective functions (a.k.a. cost functions); it allows pruning branches that will not improve the objective functions. Since monotonicity is only required for the objective function, this method can be used for any family of classifiers (e.g., linear models, neural networks, and decision trees). However, additional classifier properties speed up the tree search when it foresees branches that will not result in feasible actions. Moreover, the proposed optimization generates a diverse set of Pareto-optimal counterfactual antecedents by relying on multi-objective concepts. The results show an algorithm with working guarantees that enumerates a wide range of counterfactual antecedents. It helps the decision-maker understand the machine learning decision and finds alternatives to achieve the desired outcome. The user can inspect these multiple counterfactual antecedents to find the most suitable one and better understand the prediction. [ABSTRACT FROM AUTHOR]

The widespread deployment of deep nets in practical applications has lead to a growing desire to understand how and why such black-box methods perform prediction. Much work has focused on understanding what part of the input pattern (an image, say) is responsible for a particular class being predicted, and how the input may be manipulated to predict a different class. We focus instead on understanding which of the internal features computed by the neural net are responsible for a particular class. We achieve this by mimicking part of the neural net with an oblique decision tree having sparse weight vectors at the decision nodes. Using the recently proposed Tree Alternating Optimization (TAO) algorithm, we are able to learn trees that are both highly accurate and interpretable. Such trees can faithfully mimic the part of the neural net they replaced, and hence they can provide insights into the deep net black box. Further, we show we can easily manipulate the neural net features in order to make the net predict, or not predict, a given class, thus showing that it is possible to carry out adversarial attacks at the level of the features. These insights and manipulations apply globally to the entire training and test set, not just at a local (single-instance) level. We demonstrate this robustly in the MNIST and ImageNet datasets with LeNet5 and VGG networks. [ABSTRACT FROM AUTHOR]

In the meantime, a wide variety of terminologies, motivations, approaches, and evaluation criteria have been developed within the research field of explainable artificial intelligence (XAI). With the amount of XAI methods vastly growing, a taxonomy of methods is needed by researchers as well as practitioners: To grasp the breadth of the topic, compare methods, and to select the right XAI method based on traits required by a specific use-case context. Many taxonomies for XAI methods of varying level of detail and depth can be found in the literature. While they often have a different focus, they also exhibit many points of overlap. This paper unifies these efforts and provides a complete taxonomy of XAI methods with respect to notions present in the current state of research. In a structured literature analysis and meta-study, we identified and reviewed more than 50 of the most cited and current surveys on XAI methods, metrics, and method traits. After summarizing them in a survey of surveys, we merge terminologies and concepts of the articles into a unified structured taxonomy. Single concepts therein are illustrated by more than 50 diverse selected example methods in total, which we categorize accordingly. The taxonomy may serve both beginners, researchers, and practitioners as a reference and wide-ranging overview of XAI method traits and aspects. Hence, it provides foundations for targeted, use-case-oriented, and context-sensitive future research. [ABSTRACT FROM AUTHOR]

Predicting the capacity of lithium-ion battery (LIB) plays a crucial role in ensuring the safe operation of LIBs and prolonging their lifespan. However, LIBs are easily affected by environmental interference, which may impact the precision of predictions. Furthermore, interpretability in the process of predicting LIB capacity is also important for users to understand the model, identify issues, and make decisions. In this study, an interpretable method considering environmental interference (IM-EI) for predicting LIB capacity is introduced. Spearman correlation coefficients, interpretability principles, belief rule base (BRB), and interpretability constraints are used to improve the prediction precision and interpretability of IM-EI. Dynamic attribute reliability is introduced to minimize the effect of environmental interference. The experimental results show that IM-EI model has good interpretability and high precision compared to the other models. Under interference conditions, the model still has good precision and robustness. [ABSTRACT FROM AUTHOR]

Recurrent neural networks (RNNs) play a pivotal role in natural language processing and computer vision. Long short-term memory (LSTM), as one of the most representative RNNs, is built upon relatively complex architecture with an excessive number of parameters, which results in large storage, high training cost, and lousy interpretability. In this paper, we propose a lightweight network called Light Recurrent Unit (LRU). On the one hand, we designed an accessible gate structure, which has high interpretability and addresses the issue of gradient disappearance. On the other hand, we introduce the Stack Recurrent Cell (SRC) structure to modify the activation function, which not only expedites convergence rates but also enhances the interpretability of the network. Experimental results show that our proposed LRU has the advantages of fewer parameters, strong interpretability, and effective modeling ability for variable length sequences on several datasets. Consequently, LRU could be a promising alternative to traditional RNN models in real-time applications with space or time constraints, potentially reducing storage and training costs while maintaining high performance. [ABSTRACT FROM AUTHOR]

Facial biometrics are widely used to reliably and conveniently recognize people in photos, in videos, or from real-time webcam streams. It is therefore of fundamental importance to detect synthetic faces in images in order to reduce the vulnerability of biometrics-based security systems. Furthermore, manipulated images of faces can be intentionally shared on social media to spread fake news related to the targeted individual. This paper shows how fake face recognition models may mainly rely on the information contained in the background when dealing with generated faces, thus reducing their effectiveness. Specifically, a classifier is trained to separate fake images from real ones, using their representation in a latent space. Subsequently, the faces are segmented and the background removed, and the detection procedure is performed again, observing a significant drop in classification accuracy. Finally, an explainability tool (SHAP) is used to highlight the salient areas of the image, showing that the background and face contours crucially influence the classifier decision. [ABSTRACT FROM AUTHOR]

Safefy-critical domains often employ autonomous agents which follow a sequential decision-making setup, whereby the agent follows a policy to dictate the appropriate action at each step. AI-practitioners often employ reinforcement learning algorithms to allow an agent to find the best policy. However, sequential systems often lack clear and immediate signs of wrong actions, with consequences visible only in hindsight, making it difficult to humans to understand system failure. In reinforcement learning, this is referred to as the credit assignment problem. To effectively collaborate with an autonomous system, particularly in a safety-critical setting, explanations should enable a user to better understand the policy of the agent and predict system behavior so that users are cognizant of potential failures and these failures can be diagnosed and mitigated. However, humans are diverse and have innate biases or preferences which may enhance or impair the utility of a policy explanation of a sequential agent. Therefore, in this paper, we designed and conducted human-subjects experiment to identify the factors which influence the perceived usability with the objective usefulness of policy explanations for reinforcement learning agents in a sequential setting. Our study had two factors: the modality of policy explanation shown to the user (Tree, Text, Modified Text, and Programs) and the "first impression" of the agent, i.e., whether the user saw the agent succeed or fail in the introductory calibration video. Our findings characterize a preference-performance tradeoff wherein participants perceived languagebased policy explanations to be significantly more useable; however, participants were better able to objectively predict the agent's behavior when provided an explanation in the form of a decision tree. Our results demonstrate that user-specific factors, such as computer science experience (p < 0.05), and situational factors, such as watching agent crash (p < 0.05), can significantly impact the perception and usefulness of the explanation. This research provides key insights to alleviate prevalent issues regarding innappropriate compliance and reliance, which are exponentially more detrimental in safety-critical settings, providing a path forward for XAI developers for future work on policy-explanations. [ABSTRACT FROM AUTHOR]

Knowledge representation models that aim to present data in a structured and comprehensible manner have gained popularity as a research focus in the pursuit of achieving human-level intelligence. Humans possess the ability to understand, reason and interpret knowledge. They acquire knowledge through their experiences and utilize it to carry out various actions in the real world. Similarly, machines can also perform these tasks, a process known as knowledge representation and reasoning. In this survey, we present a thorough analysis of knowledge representation models and their crucial role in information management within the healthcare domain. We provide an overview of various models, including ontologies, first-order logic and rule-based systems. We classify four knowledge representation models based on their type, such as graphical, mathematical and other types. We compare these models based on four criteria: heterogeneity, interpretability, scalability and reasoning in order to determine the most suitable model that addresses healthcare challenges and achieves a high level of satisfaction. [ABSTRACT FROM AUTHOR]

The 2022 National Defense Strategy of the United States listed climate change as a serious threat to national security. Climate intervention methods, such as stratospheric aerosol injection, have been proposed as mitigation strategies, but the downstream effects of such actions on a complex climate system are not well understood. The development of algorithmic techniques for quantifying relationships between source and impact variables related to a climate event (i.e., a climate pathway) would help inform policy decisions. Data‐driven deep learning models have become powerful tools for modeling highly nonlinear relationships and may provide a route to characterize climate variable relationships. In this paper, we explore the use of an echo state network (ESN) for characterizing climate pathways. ESNs are a computationally efficient neural network variation designed for temporal data, and recent work proposes ESNs as a useful tool for forecasting spatiotemporal climate data. However, ESNs are noninterpretable black‐box models along with other neural networks. The lack of model transparency poses a hurdle for understanding variable relationships. We address this issue by developing feature importance methods for ESNs in the context of spatiotemporal data to quantify variable relationships captured by the model. We conduct a simulation study to assess and compare the feature importance techniques, and we demonstrate the approach on reanalysis climate data. In the climate application, we consider a time period that includes the 1991 volcanic eruption of Mount Pinatubo. This event was a significant stratospheric aerosol injection, which acts as a proxy for an anthropogenic stratospheric aerosol injection. We are able to use the proposed approach to characterize relationships between pathway variables associated with this event that agree with relationships previously identified by climate scientists. [ABSTRACT FROM AUTHOR]

Geospatial interpolation plays a pivotal role in spatial analysis because it provides high-quality data support for various spatiotemporal data mining (STDM) tasks. However, statistical methods, such as kriging, face challenges in dealing with complex geo-big data. Additionally, deep-learning-based methods, despite their exceptional performance, suffer from limitations, such as poor interpretability. To harness the complementary advantages of these statistical methods and deep learning approaches, this study proposes a novel geospatial artificial intelligence (GeoAI) framework called deep kriging neural network (DKNN). The primary contribution lies in the development of an asymmetric encoder-decoder structure, which includes a deep-learning-based spatial encoder and a geostatistics-based kriging decoder. The spatial encoder consists of three specialized neural networks, whereas the kriging decoder relies on the proposed unified kriging system. During forward propagation, the kriging decoder leverage messages from the spatial encoder to generate interpolation weights for prediction. Conversely, during backward propagation, the kriging decoder guides the spatial encoder in learning interpretable knowledge. Experiments were conducted using both synthetic and practical datasets. The results demonstrate an average improvement of 20.18% in MAE, 25.04% in RMSE and 24.06% in MAPE when compared to the best-performing baseline method. Furthermore, these results confirm the superior interpretability of our DKNN framework. [ABSTRACT FROM AUTHOR]

The explication of Convolutional Neural Networks (CNN) through xAI techniques often poses challenges in interpretation. The inherent complexity of input features, notably pixels extracted from images, engenders complex correlations. Gradient-based methodologies, exemplified by Integrated Gradients (IG), effectively demonstrate the significance of these features. Nevertheless, the conversion of these explanations into images frequently yields considerable noise. Presently, we introduce GAD (Gradient Artificial Distancing) as a supportive framework for gradient-based techniques. Its primary objective is to accentuate influential regions by establishing distinctions between classes. The essence of GAD is to limit the scope of analysis during visualization and, consequently reduce image noise. Empirical investigations involving occluded images have demonstrated that the identified regions through this methodology indeed play a pivotal role in facilitating class differentiation. • We propose GAD method to simplify explanations, providing easy-to-interpret maps. • Our gradient-based method focuses on revealing feature importance in CNNs. • GAD minimizes noise in explanations compared to usual gradient-based techniques. • Class distinction is enhanced thanks to our method. • Empirical results show that GAD locates decisive image areas for classification. [ABSTRACT FROM AUTHOR]

A new method for evaluating aircraft engine monitoring data is proposed. Commonly, prognostics and health management systems use knowledge of the degradation processes of certain engine components together with professional expert opinion to predict the Remaining Useful Life (RUL). New data-driven approaches have emerged to provide accurate diagnostics without relying on such costly processes. However, most of them lack an explanatory component to understand model learning and/or the nature of the data. A solution based on a novel recurrent version of a VAE is proposed in this paper to overcome this gap. The latent space learned by the model, trained with data from sensors placed in different parts of these engines, is exploited to build a self-explanatory map that can visually evaluate the rate of deterioration of the engines. Besides, a simple regressor model is built on top of the learned features of the encoder in order to numerically predict the RUL. As a result, remarkable prognostic accuracy is achieved, outperforming most of the novel and state-of-the-art approaches on the available modular aero-propulsion system simulation data (C-MAPSS dataset) from NASA. In addition, a practical real-world application is included for Turbofan engine data. This study shows that the proposed prognostic and explainable framework presents a promising new approach. [ABSTRACT FROM AUTHOR]

Conventional pharmacokinetic (PK) models contain several useful inductive biases guiding model convergence to more realistic predictions of drug concentrations. Implementing similar biases in standard neural networks can be challenging, but might be fundamental for model robustness and predictive performance. In this study, we build on the deep compartment model (DCM) architecture by introducing constraints that guide the model to explore more physiologically realistic solutions. Using a simulation study, we show that constraints improve robustness in sparse data settings. Additionally, predicted concentration–time curves took on more realistic shapes compared to unconstrained models. Next, we propose the use of multi-branch networks, where each covariate can be connected to specific PK parameters, to reduce the propensity of models to learn spurious effects. Another benefit of this architecture is that covariate effects are isolated, enabling model interpretability through the visualization of learned functions. We show that all models were sensitive to learning false effects when trained in the presence of unimportant covariates, indicating the importance of selecting an appropriate set of covariates to link to the PK parameters. Finally, we compared the predictive performance of the constrained models to previous relevant population PK models on a real-world data set of 69 haemophilia A patients. Here, constrained models obtained higher accuracy compared to the standard DCM, with the multi-branch network outperforming previous PK models. We conclude that physiological-based constraints can improve model robustness. We describe an interpretable architecture which aids model trust, which will be key for the adoption of machine learning-based models in clinical practice. [ABSTRACT FROM AUTHOR]

Recycled aggregate concrete is an effective solution for efficiently managing municipal construction waste on a large scale. Shear bearing capacity (SBC) is significant for reinforced concrete structures, and it is essential to develop trustworthy calculation models for structural design. This paper proposes a tree model-based SBC assessment system that considers eight design parameters of reinforced recycled concrete beams (RRCBs). Evaluation results revealed that the extreme gradient boosting model yielded the highest prediction accuracy with an R2 of 0.960 and a mean absolute percentage error of 7.343%. To reduce the risk of black-box models, this study conducted feature importance calculations, sensitivity analyses and reliability validation of the prediction results. The findings demonstrated that increasing the hoop reinforcement ratio and beam width significantly improved the SBC of RRCB. The compressive strength and longitudinal reinforcement ratio had positive effects on the SBC, while longitudinal steel bar yield strength had no effect on the SBC. These analyses can be combined with physical mechanisms to better refine the performance design. Furthermore, a comparative study utilizing two commonly used standard formulas was conducted. The results indicated that the SBCs estimated using the tree model are more accurate than those calculated by the standard formulas. [ABSTRACT FROM AUTHOR]

Although deep reinforcement learning has become a promising machine learning approach for sequential decision-making problems, it is still not mature enough for high-stake domains such as autonomous driving or medical applications. In such contexts, a learned policy needs for instance to be interpretable, so that it can be inspected before any deployment (e.g., for safety and verifiability reasons). This survey provides an overview of various approaches to achieve higher interpretability in reinforcement learning (RL). To that aim, we distinguish interpretability (as an intrinsic property of a model) and explainability (as a post-hoc operation) and discuss them in the context of RL with an emphasis on the former notion. In particular, we argue that interpretable RL may embrace different facets: interpretable inputs, interpretable (transition/reward) models, and interpretable decision-making. Based on this scheme, we summarize and analyze recent work related to interpretable RL with an emphasis on papers published in the past 10 years. We also discuss briefly some related research areas and point to some potential promising research directions, notably related to the recent development of foundation models (e.g., large language models, RL from human feedback). [ABSTRACT FROM AUTHOR]

Counterfactual explanations modify the feature values of an instance in order to alter its prediction from an undesired to a desired label. As such, they are highly useful for providing trustworthy interpretations of decision-making in domains where complex and opaque machine learning algorithms are utilized. To guarantee their quality and promote user trust, they need to satisfy the faithfulness desideratum, when supported by the data distribution. We hereby propose a counterfactual generation algorithm for mixed-feature spaces that prioritizes faithfulness through k-justification, a novel counterfactual property introduced in this paper. The proposed algorithm employs a graph representation of the search space and provides counterfactuals by solving an integer program. In addition, the algorithm is classifier-agnostic and is not dependent on the order in which the feature space is explored. In our empirical evaluation, we demonstrate that it guarantees k-justification while showing comparable performance to state-of-the-art methods in feasibility, sparsity, and proximity. [ABSTRACT FROM AUTHOR]

Packing problems have been studied for a long time and have great applications in real-world scenarios. In recent times, with problems in the industrial world increasing in size, exact algorithms are often not a viable option and faster approaches are needed. We study Monte Carlo tree search, a random sampling algorithm that has gained great importance in literature in the last few years. We propose three approaches based on MCTS and its integration with metaheuristic algorithms or deep learning models to obtain approximated solutions to packing problems that are also interpretable by means of MCTS exploration and from which knowledge can be extracted. We focus on two-dimensional rectangle packing problems in our experimentation and use several well known benchmarks from literature to compare our solutions with existing approaches and offer a view on the potential uses for knowledge extraction from our method. We manage to match the quality of state-of-the-art methods, with improvements in time with respect to some of them and greater interpretability. [ABSTRACT FROM AUTHOR]

Many applications are driven by Machine Learning (ML) today. While complex ML models lead to an accurate prediction, their inner decision-making is obfuscated. However, especially for high-stakes decisions, interpretability and explainability of the model are necessary. Therefore, we develop a holistic interpretability and explainability framework (HIEF) to objectively describe and evaluate an intelligent system's explainable AI (XAI) capacities. This guides data scientists to create more transparent models. To evaluate our framework, we analyse 50 real estate appraisal papers to ensure the robustness of HIEF. Additionally, we identify six typical types of intelligent systems, so-called archetypes, which range from explanatory to predictive, and demonstrate how researchers can use the framework to identify blind-spot topics in their domain. Finally, regarding comprehensiveness, we used a random sample of six intelligent systems and conducted an applicability check to provide external validity. [ABSTRACT FROM AUTHOR]

Objective: To construct a risk prediction model for assisted diagnosis of Diabetic Nephropathy (DN) using machine learning algorithms, and to validate it internally and externally. Methods: Firstly, the data was cleaned and enhanced, and was divided into training and test sets according to the 7:3 ratio. Then, the metrics related to DN were filtered by difference analysis, Least Absolute Shrinkage and Selection Operator (LASSO), Recursive Feature Elimination (RFE), and Max-relevance and Min-redundancy (MRMR) algorithms. Ten machine learning models were constructed based on the key variables. The best model was filtered by Receiver Operating Characteristic (ROC), Precision-Recall (PR), Accuracy, Matthews Correlation Coefficient (MCC), and Kappa, and was internally and externally validated. Based on the best model, an online platform had been constructed. Results: 15 key variables were selected, and among the 10 machine learning models, the Random Forest model achieved the best predictive performance. In the test set, the area under the ROC curve was 0.912, and in two external validation cohorts, the area under the ROC curve was 0.828 and 0.863, indicating excellent predictive and generalization abilities. Conclusion: The model has a good predictive value and is expected to help in the early diagnosis and screening of clinical DN. [ABSTRACT FROM AUTHOR]

As the range of decisions made by Artificial Intelligence (AI) expands, the need for Explainable AI (XAI) becomes increasingly critical. The reasoning behind the specific outcomes of complex and opaque financial models requires a thorough justification to improve risk assessment, minimise the loss of trust, and promote a more resilient and trustworthy financial ecosystem. This Systematic Literature Review (SLR) identifies 138 relevant articles from 2005 to 2022 and highlights empirical examples demonstrating XAI's potential benefits in the financial industry. We classified the articles according to the financial tasks addressed by AI using XAI, the variation in XAI methods between applications and tasks, and the development and application of new XAI methods. The most popular financial tasks addressed by the AI using XAI were credit management, stock price predictions, and fraud detection. The three most commonly employed AI black-box techniques in finance whose explainability was evaluated were Artificial Neural Networks (ANN), Extreme Gradient Boosting (XGBoost), and Random Forest. Most of the examined publications utilise feature importance, Shapley additive explanations (SHAP), and rule-based methods. In addition, they employ explainability frameworks that integrate multiple XAI techniques. We also concisely define the existing challenges, requirements, and unresolved issues in applying XAI in the financial sector. [ABSTRACT FROM AUTHOR]

Drug repurposing aims to find new therapeutic applications for existing drugs in the pharmaceutical market, leading to significant savings in time and cost. The use of artificial intelligence and knowledge graphs to propose repurposing candidates facilitates the process, as large amounts of data can be processed. However, it is important to pay attention to the explainability needed to validate the predictions. We propose a general architecture to understand several explainable methods for graph completion based on knowledge graphs and design our own architecture for drug repurposing. We present XG4Repo (eXplainable Graphs for Repurposing), a framework that takes advantage of the connectivity of any biomedical knowledge graph to link compounds to the diseases they can treat. Our method allows methapaths of different types and lengths, which are automatically generated and optimised based on data. XG4Repo focuses on providing meaningful explanations to the predictions, which are based on paths from compounds to diseases. These paths include nodes such as genes, pathways, side effects, or anatomies, so they provide information about the targets and other characteristics of the biomedical mechanism that link compounds and diseases. Paths make predictions interpretable for experts who can validate them and use them in further research on drug repurposing. We also describe three use cases where we analyse new uses for Epirubicin, Paclitaxel, and Predinisone and present the paths that support the predictions. [ABSTRACT FROM AUTHOR]

Predicting student performance in the future is a crucial behavior prediction problem in education. By predicting student performance, educational experts can provide individualized instruction, optimize the allocation of resources, and develop educational strategies. If the prediction results are unreliable, it is difficult to earn the trust of educational experts. Therefore, prediction methods need to satisfy the requirement of interpretability. For this reason, the prediction model is constructed in this paper using belief rule base (BRB). BRB not only combines expert knowledge, but also has good interpretability. There are two problems in applying BRB to student performance prediction: first, in the modeling process, the system is too complex due to the large number of indicators involved. Secondly, the interpretability of the model can be compromised during the optimization process. To overcome these challenges, this paper introduces a hierarchical belief rule base with interpretability (HBRB-I) for student performance prediction. First, it analyzes how the HBRB-I model achieves interpretability. Then, an attribute grouping method is proposed to construct a hierarchical structure by reasonably organizing the indicators, so as to effectively reduce the complexity of the model. Finally, an objective function considering interpretability is designed and the projected covariance matrix adaptive evolution strategy (P-CMA-ES) optimization algorithm is improved. The aim is to ensure that the model remains interpretable after optimization. By conducting experiments on the student performance dataset, it is demonstrated that the proposed model performs well in terms of both accuracy and interpretability. [ABSTRACT FROM AUTHOR]

Recently, various deep learning methods have been applied to Modulation Format Identification (MFI). The interpretability of deep learning models is important. However, this interpretability is challenged due to the black-box nature of deep learning. To deal with this difficulty, we propose an Adaptive Mask-Based Interpretable Convolutional Neural Network (AMI-CNN) that utilizes a mask structure for feature selection during neural network training and feeds the selected features into the classifier for decision making. During training, the masks are updated dynamically with parameters to optimize feature selection. The extracted mask serves as interpretable weights, with each weight corresponding to a feature, reflecting the contribution of each feature to the model's decision. We validate the model on two datasets—Power Spectral Density (PSD) and constellation phase histogram—and compare it with three classical interpretable methods: Gradient-Weighted Class Activation Mapping (Grad-CAM), Local Interpretable Model-Agnostic Explanations (LIME), and Shapley Additive exPlanations (SHAP). The MSE values are as follows: AMI-CNN achieves the lowest MSE of 0.0246, followed by SHAP with 0.0547, LIME with 0.0775, and Grad-CAM with 0.1995. Additionally, AMI-CNN achieves the highest PG-Acc of 1, whether on PSD or on constellation phase histogram. Experimental results demonstrate that the AMI-CNN model outperforms compared methods in both qualitative and quantitative analyses. [ABSTRACT FROM AUTHOR]

Although anion‐exchange membranes (AEMs) are commonly used in fuel cells and water electrolyzers, their widespread commercialization is hindered by problems such as low anion conductivity and durability. Moreover, the development of high‐performance AEMs remains complex and time consuming. Here, we address these challenges by proposing an innovative approach for the efficient design and screening of AEM polymers using unsupervised machine learning. Our model, which combines principal component analysis with uniform manifold approximation and projection, generates an intuitive map that clusters AEM polymers based on structural similarities without any predefined knowledge regarding anion conductivity or other experimentally derived variables. As a powerful navigation tool, this map provides insights into promising main‐chain structures, such as poly(arylene alkylene)s with consistently high conductivity and polyolefins with exceptional performance depending on the substituent. Furthermore, assisted by key molecular descriptors, inverse analysis with this model allows targeted design and property prediction before synthesis, which will significantly accelerate the discovery of novel AEM polymers. This work represents a paradigm shift not only in AEM research but also generally in materials research, moving from black‐box predictions toward interpretable guidelines that foster collaboration between researchers and machine learning for efficient and informed material development. [ABSTRACT FROM AUTHOR]

Mechanical equipment is composed of several parts, and the interaction between parts exists throughout the whole life cycle, leading to the widespread phenomenon of fault coupling. The diagnosis of independent faults cannot meet the requirements of the health management of mechanical equipment under actual working conditions. In this paper, the dynamic vertex interpretable graph neural network (DIGNN) is proposed to solve the problem of coupling fault diagnosis, in which dynamic vertices are defined in the data topology. First, in the date preprocessing phase, wavelet transform is utilized to make input features interpretable and reduce the uncertainty of model training. In the fault topology, edge connections are made between nodes according to the fault coupling information, and edge connections are established between dynamic nodes and all other nodes. Second the data topology with dynamic vertices is used in the training phase and in the testing phase, the time series data are only fed into dynamic vertices for classification and analysis, which makes it possible to realize coupling fault diagnosis in an industrial production environment. The features extracted in different layers of DIGNN interpret how the model works. The method proposed in this paper can realize the accurate diagnosis of independent faults in the dataset with an accuracy of 100%, and can effectively judge the coupling mode of coupling faults with a comprehensive accuracy of 88.3%. [ABSTRACT FROM AUTHOR]

In this study, we present a novel approach to enhancing the interpretability of medical image classification by integrating formal concept analysis (FCA) with convolutional neural networks (CNNs). While CNNs are increasingly applied in medical diagnoses, understanding their decision-making remains a challenge. Although visualization techniques like saliency maps offer insights into CNNs' decision-making for individual images, they do not explicitly establish a relationship between the high-level features learned by CNNs and the class labels across entire dataset. To bridge this gap, we leverage the FCA framework as an image classification model, presenting a novel method for understanding the relationship between abstract features and class labels in medical imaging. Building on our previous work, which applied this method to the MNIST handwritten image dataset and demonstrated that the performance is comparable to CNNs, we extend our approach and evaluation to histopathological image datasets, including Warwick-QU and BreakHIS. Our results show that the FCA-based classifier offers comparable accuracy to deep neural classifiers while providing transparency into the classification process, an important factor in clinical decision-making. [ABSTRACT FROM AUTHOR]

Climate change is causing permafrost in the Qinghai–Tibet Plateau to degrade, triggering thermokarst hazards and impacting the environment. Despite their ecological importance, the distribution and risks of thermokarst lakes are not well understood due to complex influencing factors. In this study, we introduced a new interpretable ensemble learning method designed to improve the global and local interpretation of susceptibility assessments for thermokarst lakes. Our primary aim was to offer scientific support for precisely evaluating areas prone to thermokarst lake formation. In the thermokarst lake susceptibility assessment, we identified ten conditioning factors related to the formation and distribution of thermokarst lakes. In this highly accurate stacking model, the primary learning units were the random forest (RF), extremely randomized trees (EXTs), extreme gradient boosting (XGBoost), and categorical boosting (CatBoost) algorithms. Meanwhile, gradient boosted decision trees (GBDTs) were employed as the secondary learning unit. Based on the stacking model, we assessed thermokarst lake susceptibility and validated accuracy through six evaluation indices. We examined the interpretability of the stacking model using three interpretation methods: accumulated local effects (ALE), local interpretable model-agnostic explanations (LIME), and Shapley additive explanations (SHAP). The results showed that the ensemble learning stacking model demonstrated superior performance and the highest prediction accuracy. Approximately 91.20% of the total thermokarst hazard points fell within the high and very high susceptible areas, encompassing 20.08% of the permafrost expanse in the QTP. The conclusive findings revealed that slope, elevation, the topographic wetness index (TWI), and precipitation were the primary factors influencing the assessment of thermokarst lake susceptibility. This comprehensive analysis extends to the broader impacts of thermokarst hazards, with the identified high and very high susceptibility zones affecting significant stretches of railway and highway infrastructure, substantial soil organic carbon reserves, and vast alpine grasslands. This interpretable ensemble learning model, which exhibits high accuracy, offers substantial practical significance for project route selection, construction, and operation in the QTP. [ABSTRACT FROM AUTHOR]

This study introduces an intelligent method for regional subsurface prediction using a Stacking ensemble learning approach, which incorporates K-Nearest Neighbors (KNN), Decision Tree (DT), Random Forest (RF), Gradient Boosted Decision Trees (GBDT), and Xgboost as base classifiers, with Logistic Regression (LR) serving as the meta-classifier. Leveraging data from 1119 boreholes in Zigong City, China, this method achieves a prediction accuracy of 93%, and notably improves the prediction of weak layers, with accuracy rates ranging from 71.4% to 81.5%. This enhancement is particularly significant in areas with a random distribution of excavation and backfill. Furthermore, this study employs the SHAP method (SHapley Additive explanations) to interpret the Stacking ensemble learning model, revealing that the outputs of the base classifiers enhance the feature set for the meta-classifier, effectively addressing the insensitivity of the spatial coordinates x, y, and z as input features for lithology prediction. The findings demonstrate that the expansion of effective feature dimensions is key to the superior performance of the Stacking ensemble learning method in regional subsurface lithology prediction. [ABSTRACT FROM AUTHOR]

In this paper we study essential hereditary undecidability. Theories with this property are a convenient tool to prove undecidability of other theories. The paper develops the basic facts concerning essentially hereditary undecidability and provides salient examples, like a construction of essentially hereditarily undecidable theories due to Hanf and an example of a rather natural essentially hereditarily undecidable theory strictly below R. We discuss the (non-)interaction of essential hereditary undecidability with recursive boolean isomorphism. We develop a reduction relation essential tolerance, or, in the converse direction, lax interpretability that interacts in a good way with essential hereditary undecidability. We introduce the class of Σ 1 0 -friendly theories and show that Σ 1 0 -friendliness is sufficient but not necessary for essential hereditary undecidability. Finally, we adapt an argument due to Pakhomov, Murwanashyaka and Visser to show that there is no interpretability minimal essentially hereditarily undecidable theory. [ABSTRACT FROM AUTHOR]

The goal of this study is to reconstruct a high-quality computed tomography (CT) image from low-dose acquisition using an unrolling deep learning-based reconstruction network with less computational complexity and a more generalized model. We propose a MDUNet: Multi-parameters deep-prior unrolling network that employs the cascaded convolutional and deconvolutional blocks to unroll the model-based iterative reconstruction within a finite number of iterations by data-driven training. Furthermore, the embedded data consistency constraint in MDUNet ensures that the input low-dose images and the low-dose sinograms are consistent as well as incorporate the physics imaging geometry. Additionally, multi-parameter training was employed to enhance the model's generalization during the training process. Experimental results based on AAPM Low-dose CT datasets show that the proposed MDUNet significantly outperforms other state-of-the-art (SOTA) methods quantitatively and qualitatively. Also, the cascaded blocks reduce the computational complexity with reduced training parameters and generalize well on different datasets. In addition, the proposed MDUNet is validated on 8 different organs of interest, with more detailed structures recovered and high-quality images generated. The experimental results demonstrate that the proposed MDUNet generates favorable improvement over other competing methods in terms of visual quality, quantitative performance, and computational efficiency. The MDUNet has improved image quality with reduced computational cost and good generalization which effectively lowers radiation dose and reduces scanning time, making it favorable for future clinical deployment. [ABSTRACT FROM AUTHOR]

Recently, the physics-informed neural network (PINN) has attracted much attention in solving partial differential equations (PDEs). The success is due to the strong generalization ability of the neural network (NN), which is supported by the universal approximation theorem, and its mesh-free implementation. In this paper, we propose a multi-head NN enriched PINN (MultiPINN) for solving differential equations. The trial function is built based on the radial basis function (RBF)-interpolation, which makes NN training parameters partially interpre. The loss function is constructed by embedding the physics information of differential equations and boundary conditions. Then the parameters in MultiPINN are trained using the ADAM optimizer. A significant feature of MultiPINN is that it combines the traditional RBF interpolation method with machine learning (ML) techniques. The ML technique is employed to learn the basis feature enrichment that provides global information. The multi-head mechanism is used so that each node has multiple bases, which can improve the accuracy of the MultiPINN solution. Two ordinary differential equations and three partial differential equations, i.e. the convection equation, the Burgers equation, and the Poisson equation, are used in the numerical experiments. The experimental outcomes demonstrate that MultiPINN produces solutions consistent with both analytical solutions and solutions obtained through traditional numerical methods. Additionally, MultiPINN shows robustness and adaptability over the other NN-based methods in the implementations. [ABSTRACT FROM AUTHOR]

This research presents a method to analyze how neural network models, applied to Expanded Polypropylene and Expanded Polystyrene foams, predict their compressive stress responses. By using SHAP values and Partial Dependence Plots, the study elucidates the models' decision-making processes. It focuses on three main features for both materials: density, loading rate, and strain, with an additional feature concerning loading and unloading for Expanded Polystyrene foam. The findings highlight that increased density and loading rate are closely correlated with higher compressive responses, and strain emerges as the most influential factor for the response of both materials. Partial Dependence Plots reveal a linear relationship with density, whereas other variables demonstrate non-linear relationships. These results validate the use of neural networks in analyzing material behavior, showing that the models' outputs are in line with empirical observations. In conclusion, as presented, the integration of interpretability tools with neural network models offers a robust method for material response analysis, contributing to a deeper understanding of material science. [ABSTRACT FROM AUTHOR]

Sentiment analysis can be used to derive knowledge that is connected to emotions and opinions from textual data generated by people. As computer power has grown, and the availability of benchmark datasets has increased, deep learning models based on deep neural networks have emerged as the dominant approach for sentiment analysis. While these models offer significant advantages, their lack of interpretability poses a major challenge in comprehending the rationale behind their reasoning and prediction processes, leading to complications in the models' explainability. Further, only limited research has been carried out into developing deep learning models that describe their internal functionality and behaviors. In this timely study, we carry out a first of its kind overview of key sentiment analysis techniques and eXplainable artificial intelligence (XAI) methodologies that are currently in use. Furthermore, we provide a comprehensive review of sentiment analysis explainability. [ABSTRACT FROM AUTHOR]

Obstructive Sleep Apnea (OSA) is a prevalent sleep-related breathing disorder that leads to various health issues such as hypertension, heart disease, diabetes, and stroke. In order to achieve a convenient, robust and accurate OSA detection, we analyze the cardiopulmonary coupling mechanism of OSA from single-lead electrocardiogram (ECG) signals. Then we propose a fine-grained convolutional recurrent model (FCRM) for obstructive sleep apnea detection to learn the variation of cardiopulmonary coupling (CPC) features for OSA detection. Finally, we offer interpretable insights into the model's decisions using respiration signal and achieve fine-grained apnea classification based on attention score. The proposed model's performance on the Apnea-ECG dataset achieved 93.2% accuracy, 89.2% sensitivity, and 96.4% specificity. This demonstrates that the method effectively extracts cardiopulmonary characteristics during sleep apnea and outperforms other methods. [ABSTRACT FROM AUTHOR]