Enhancing Transparency and Interpretability in Deep Learning Models: A Comprehensive Study on Explainable AI Techniques

The rapid proliferation of artificial intelligence (AI) technologies across various industries and decision-making processes has undeniably transformed the way of approaching complex problems and tasks. AI systems have proven their prowess in areas such as healthcare, finance, and autonomous systems, revolutionizing how decisions are made. Nevertheless, this proliferation of AI has raised critical concerns regarding the transparency, accountability, and fairness of these systems, as many of the state-of-the-art AI models often resemble complex black boxes. These intricate models, particularly deep learning neural networks, harbor non-linear relationships that are difficult for human users to decipher, thereby raising concerns about bias, fairness, and overall trustworthiness in AI-driven decisions. The urgency of this issue is underscored by the realization that AI should not merely be accurate; it should also be interpretable. Explainable AI (XAI) has emerged as a vital field of research, emphasizing the development of models and techniques that render AI systems comprehensible and transparent in their decision-making processes. This paper investigates into the relevance and significance of XAI across various domains, including healthcare, finance, and autonomous systems, where the ability to understand the rationale behind AI decisions is paramount. In healthcare, where AI assists in diagnosis and treatment, the interpretability of AI models is crucial for clinicians to make informed decisions. In finance, applications like credit scoring and investment analysis demand transparent AI to ensure fairness and accountability. In the realm of autonomous systems, transparency is indispensable to guarantee safety and compliance with regulations. Moreover, government agencies in areas such as law enforcement and social services require interpretable AI to maintain ethical standards and accountability. This paper also highlights the diverse array of research efforts in the XAI domain, spanning from model-specific interpretability methods to more general approaches aimed at unveiling complex AI models. Interpretable models like decision trees and rule-based systems have gained attention for their inherent transparency, while integrating explanation layers into deep neural networks strives to balance accuracy with interpretability. The study emphasizes the significance of this burgeoning field in bridging the gap between AI's advanced capabilities and human users' need for comprehensible AI systems. It seeks to contribute to this field by exploring the design, development, and practical applications of interpretable AI models and techniques, with the ultimate goal of enhancing the trust and understanding of AI-driven decisions.

Text classification plays an essential role in the medical domain by organizing and categorizing vast amounts of textual data through machine learning (ML) and deep learning (DL). The adoption of Artificial Intelligence (AI) technologies in healthcare has raised concerns about the interpretability of AI models, often perceived as "black boxes." Explainable AI (XAI) techniques aim to mitigate this issue by elucidating AI model decision-making process. In this paper, we present a scoping review exploring the application of different XAI techniques in medical text classification, identifying two main types: model-specific and model-agnostic methods. Despite some positive feedback from developers, formal evaluations with medical end users of these techniques remain limited. The review highlights the necessity for further research in XAI to enhance trust and transparency in AI-driven decision-making processes in healthcare.

The accelerating complexity of global regulatory frameworks, driven by rapid technological advancements, cross-border transactions, and shifting socio-economic priorities, has placed unprecedented demands on organizations to maintain continuous compliance. Traditional compliance management systems, often rule-based and manually updated, struggle to adapt to the dynamic and fragmented nature of these evolving landscapes. This paper proposes the design of intelligent compliance systems that leverage artificial intelligence (AI), machine learning (ML), and natural language processing (NLP) to automate regulatory monitoring, interpretation, and enforcement. By integrating real-time data streams from multiple jurisdictions, the proposed system employs semantic analysis to extract, classify, and map regulatory requirements to organizational policies, operational processes, and risk controls. A modular architecture is developed to ensure scalability, interoperability, and adaptability, enabling sector-specific customization and rapid incorporation of regulatory changes. The system incorporates predictive analytics to forecast regulatory trends, simulate compliance scenarios, and recommend proactive adjustments, thereby transforming compliance from a reactive obligation into a strategic advantage. Furthermore, explainable AI techniques are embedded to enhance transparency and trust, ensuring that automated decisions align with both legal mandates and ethical standards. Case studies across finance, healthcare, and energy sectors illustrate how intelligent compliance systems reduce operational risk, lower compliance costs, and improve audit readiness. The research underscores the importance of harmonizing technological innovation with robust governance frameworks to mitigate algorithmic bias, protect sensitive data, and meet jurisdiction-specific legal obligations such as GDPR, CCPA, and sectoral regulations. This work concludes that intelligent compliance systems represent a paradigm shift, enabling organizations to navigate the evolving global regulatory landscape with agility, accuracy, and strategic foresight, while fostering regulatory harmonization and operational resilience in an increasingly interconnected world. Keywords: Intelligent Compliance Systems, Artificial Intelligence, Machine Learning, Regulatory Technology, RegTech, Global Regulations, Natural Language Processing, Predictive Analytics, Explainable AI, Compliance Automation, Governance Frameworks, Operational Resilience, Risk Management, Legal Technology, Regulatory Harmonization.

Money laundering has been a global issue for decades, which is one of the major threat for economy and society. Government, regulatory and financial institutions are combating it together in their respective capacity, however still billions of dollars in fines by authorities make the headlines in the news. High-speed internet services have enabled financial institutions to deliver better customer experience through multi-channel engagements, which has led to exponential growth in transactions and new avenues for laundering the money for fraudsters. Literature shows the usage of statistical methods, data mining and Machine Learning (ML) techniques for money laundering detection, but limited research on Deep Learning (DL) techniques, primarily due to lack of model interpretability and explainability of the decisions made. Several studies are conducted on application of ML for Anti-Money Laundering (AML), and Explainable Artificial Intelligence (XAI) techniques in general, but lacks the study on usage of DL techniques together with XAI. This paper aims to review the current state-of-the-art literature on DL together with XAI for identifying suspicious money laundering transactions and identify future research areas. Key findings of the review are, researchers have preferred variants of Convolutional Neural Networks, and AutoEncoder; graph deep learning together with natural language processing is emerging as an important technology for AML; XAI use is not seen in AML domain; 51% ML methods used in AML are non-interpretable, 58% studies used sample of old real data; key challenges for researchers are access to recent real transaction data and scarcity of labelled training data; and data being highly imbalanced. Future research directions are, application of XAI techniques to bring-out explainability, graph deep learning using natural language processing (NLP), unsupervised and reinforcement learning to handle lack of labelled data; and joint research programs between research community and industry to benefit from domain knowledge and controlled access to data.

Non-small cell lung cancer (NSCLC) remains a major global health challenge, with high post-surgical recurrence rates underscoring the need for accurate pathological response predictions to guide personalized treatments. Although artificial intelligence models show promise in this domain, their clinical adoption is limited by the lack of medically grounded guidance during training, often resulting in non-explainable intrinsic predictions. To address this, we propose Doctor-in-the-Loop , a novel framework that integrates expert-driven domain knowledge with explainable artificial intelligence techniques, directing the model toward clinically relevant anatomical regions and improving both interpretability and trustworthiness. Our approach employs a gradual multi-view strategy, progressively refining the model’s focus from broad contextual features to finer, lesion-specific details. By incorporating domain insights at every stage, we enhance predictive accuracy while ensuring that the model’s decision-making process aligns more closely with clinical reasoning. Evaluated on a dataset of NSCLC patients, Doctor-in-the-Loop delivers promising predictive performance and provides transparent, justifiable outputs, representing a significant step toward clinically explainable artificial intelligence in oncology. • Integrating clinical expertise with XAI for NSCLC pathological response prediction. • Gradual multi-view strategy refines focus from context to lesion-specific regions. • Improved explainability guides the model to clinically relevant anatomical regions. • Outperforms state-of-the-art, achieving higher accuracy and clinical relevance.

This study presents a comprehensive evaluation of state-of-the-art artificial intelligence (AI) techniques for brain tumor classification using magnetic resonance imaging (MRI). Leveraging publicly available datasets comprising over 7,000 T1-weighted contrast-enhanced scans across glioma, meningioma, pituitary tumor, and normal classes, we implemented rigorous preprocessing, including noise reduction, skull stripping, normalization, and contrast enhancement. Multiple deep learning architectures, including VGG, ResNet, DenseNet, EfficientNet, and custom convolutional neural networks (CNNs), were trained using transfer learning and fine-tuning strategies. Hybrid and ensemble models combining CNNs with Long Short-Term Memory (LSTM) networks and optimization algorithms further improved diagnostic accuracy. Our best-performing model, EfficientNetB4 with Adam optimization and targeted data augmentation, achieved a classification accuracy of 99.66% and a perfect F1-score, demonstrating clinical-grade performance. Explainable AI techniques such as Grad-CAM, LIME, and SHAP were employed to enhance model interpretability and foster clinical trust. The study highlights the potential of AI-driven diagnostics to surpass traditional methods by offering rapid, objective, and highly accurate tumor classification, facilitating personalized treatment planning. Challenges related to dataset heterogeneity, generalizability, and ethical deployment are discussed, along with future directions involving multimodal data integration and federated learning. These advancements underscore AI’s transformative role in neuro-oncology diagnostics.

Artificial intelligence (AI) has profoundly altered the financial services industry by offering new paradigms for data analysis, decision-making, and operational efficiency. This chapter provides an extensive review of the core concepts, recent advancements, and emerging applications of AI in the financial sector. Based on a critical review of 162 high-impact scientific articles, the chapter identifies and evaluates key theme areas, including financial forecasting, fraud detection, explainable AI (XAI), and the nexus between AI and FinTech. The findings demonstrate the importance of state-of-the-art AI techniques such as deep learning, reinforcement learning, and hybrid models for enhancing risk assessment, raising forecast precision, and ensuring regulatory compliance. The chapter also discusses how AI could transform several financial processes, including portfolio management, credit assessment, and anti-money laundering efforts. The creation of interpretable AI frameworks, the incorporation of multimodal data sources, and the examination of AI's societal ramifications regarding sustainability and financial inclusion are just a few of the new opportunities and research gaps that are highlighted in this chapter. By bridging the gap between academic research and industry practice, this chapter provides useful insights for both researchers and practitioners. To harness AI's transformative potential while upholding moral standards and public trust, it encourages ethical, interdisciplinary collaboration. The chapter concludes with a strategic roadmap for the innovative and long-term use of AI technology in contemporary financial ecosystems.

This work focuses on applying explainable artificial intelligence (AI) techniques to improve the interpretability and reliability of models in two crucial scenarios: detecting collusion in public procurement auctions and identifying corruption in public contracts in Mexico. While these challenges are specific, the need for transparency and explainability in AI systems is universal, especially when it comes to high-impact government issues. Our analysis is based on case studies of these particular applications, highlighting how interpretability techniques such as SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations) can be used to provide crucial insights into decisions of the algorithms involved. In doing so, our paper seeks to contribute to a broader understanding of how explainable AI can be applied in diverse government contexts. Additionally, we explore the ethical and practical implications of explainable AI in governance, highlighting its ability to promote greater transparency, accountability, and trust in automated decisions. By the end of this study, we hope to provide a more comprehensive view on how explainable AI can be a valuable tool in improving public governance, not only in relation to cases of collusion and corruption, but also in a wide range of critical government applications.

The use of artificial intelligence (AI) systems is significantly increased in the past few years. AI system is expected to provide accurate predictions and it is also crucial that the decisions made by the AI systems are humanly interpretable i.e. anyone must be able to understand and comprehend the results produced by the AI system. AI systems are being implemented even for simple decision support and are easily accessible to the common man on the tip of their fingers. The increase in usage of AI has come with its own limitation, i.e. its interpretability. This work contributes towards the use of explainability methods such as local interpretable model-agnostic explanations (LIME) to interpret the results of various black box models. The conclusion is that, the bidirectional long short-term memory (LSTM) model is superior for sentiment analysis. The operations of a random forest classifier, a black box model, using explainable artificial intelligence (XAI) techniques like LIME is used in this work. The features used by the random forest model for classification are not entirely correct. The use of LIME made this possible. The proposed model can be used to enhance performance, which raises the trustworthiness and legitimacy of AI systems.

As Artificial Intelligence (AI) systems become increasingly integrated into educational settings, the demand for transparency and trustworthiness has grown. Natural Language Processing (NLP)-powered applications such as intelligent tutoring systems, automated essay scoring, and educational chatbots offer significant benefits for personalized learning, yet often operate as “black boxes.” The lack of explainability in these models can undermine user trust, raise ethical concerns, and limit their effective use in classrooms. Explainable Artificial Intelligence (XAI) offers a critical solution by making AI decisions interpretable and justifiable to end-users. This review explores the role of XAI in enhancing transparency and trust within NLP-powered educational systems. It examines core challenges faced by educators and learners when using opaque AI, including bias, accountability, and adoption resistance. The paper reviews XAI techniques such as feature attribution, attention visualization, and open learner models that provide insights into model behavior. Real-world applications like the iRead literacy tutor and AI chatbots for feedback analysis illustrate how XAI can improve stakeholder confidence and system usability. The paper also outlines future research directions, emphasizing the need for user-centered explanations, multimodal transparency, and standardized evaluation frameworks. Ultimately, the integration of XAI into educational NLP tools is not merely a technical enhancement—it is essential for building ethical, effective, and human-aligned AI systems in education. By making AI outputs understandable and actionable, XAI bridges the gap between powerful algorithms and pedagogical trust.

Credit risk assessment has become one of the major concerns in modern finance regarding informed lending decisions. Although several studies have used traditional logistic regression and linear discriminant analysis techniques, these have increasingly become inadequate tools in today’s complex and data-rich environment. Such models often struggle with large datasets and nonlinear relationships, thus reducing their predictive power and adaptability. Artificial Intelligence (AI) and Machine Learning (ML) provide two of the most innovative approaches to credit risk modeling. This paper reviews a few ML models applied to improve the accuracy and efficiency of credit risk assessment, from Random Forests and Support Vector Machines to Neural Networks. Compared to the more traditional models, AI models can enhance predictive accuracy by using a wealth of structured and unstructured information, including alternative information sources such as social media activities and transaction history. However, despite noticeable advantages, there are some challenges concerning the use of AI in credit risk assessment, including model opaqueness, bias, and regulatory compliance. The nature of such a “black box”, especially for deep learning algorithms, can limit their interpretability and complicate regulatory compliance and decision rationalization. To solve problems induced by this “black box” nature, explainable AI techniques, namely Shapley values and LIME, have been implemented to enhance the transparency of models and raise stakeholder trust in support systems for decision-making. This review aims to evaluate the current applications of AI and ML in credit risk assessment, weigh the strengths and limitations of various models, and discuss the ethical considerations and regulatory challenges linked to their adoption by credit institutions.

The recent increase of artificial intelligence (AI) systems has led to several questions, including “Can AI assessments or predictions be trusted?” and “How can people be encouraged to accept AI-based recommendations?” These questions have led to the rise of explainable AI. This short talk presents a scenario in which AI could provide information about the seabed to a sonar operator. In this scenario, received ship noise is input to a deep learning model trained to predict a seabed class. The predicted seabed class is then used to calculate transmission loss (TL) as a function of range. These TL curves from the predicted seabed class are compared to those obtained using the seabed information stored in a database. This comparison could allow the sonar operator to evaluate the applicability of the seabed database for their present location. Adoption of this type of AI tool depends on the attitude of the sonar operator about the AI and its predictions. Ideas are given for how this attitude could be improved by adoption of explainable AI techniques. [Work supported by the Office of Naval Research grant #N00014-22-12402.]

In recent years, there has been a surge of interest in Artificial Intelligence (AI) systems that can provide human-centric explanations for decisions or predictions. No matter how good and efficient an AI model is, users or practitioners find it difficult to trust it if they cannot understand the AI model or its behaviours. Incorporating explainability that is human-centric in event detection systems is significant for building a decision-making process that is more trustworthy and sustainable. Human-centric and semantics-based explainable event detection will achieve trustworthiness, explainability, and reliability, which are currently lacking in AI systems. This paper provides a survey on human-centric explainable AI, explainable event detection, and semantics-based explainable event detection by answering some research questions that bother on the characteristics of human-centric explanations, the state of explainable AI, methods for human-centric explanations, the essence of human-centricity in explainable event detection, research efforts in explainable event solutions, and the benefits of integrating semantics into explainable event detection. The findings from the survey show the current state of human-centric explainability, the potential of integrating semantics into explainable AI, the open problems, and the future directions which can guide researchers in the explainable AI domain.

Skin diseases impact millions of people around the world and pose a severe risk to public health. These diseases have a wide range of effects on the skin’s structure, functionality, and appearance. Identifying and predicting skin diseases are laborious processes that require a complete physical examination, a review of the patient’s medical history, and proper laboratory diagnostic testing. Additionally, it necessitates a significant number of histological and clinical characteristics for examination and subsequent treatment. As a disease’s complexity and quantity of features grow, identifying and predicting it becomes more challenging. This research proposes a deep learning (DL) model utilizing transfer learning (TL) to quickly identify skin diseases like chickenpox, measles, and monkeypox. A pre-trained VGG16 is used for transfer learning. The VGG16 can identify and predict diseases more quickly by learning symptom patterns. Images of the skin from the four classes of chickenpox, measles, monkeypox, and normal are included in the dataset. The dataset is separated into training and testing. The experimental results performed on the dataset demonstrate that the VGG16 model can identify and predict skin diseases with 93.29% testing accuracy. However, the VGG16 model does not explain why and how the system operates because deep learning models are black boxes. Deep learning models’ opacity stands in the way of their widespread application in the healthcare sector. In order to make this a valuable system for the health sector, this article employs layer-wise relevance propagation (LRP) to determine the relevance scores of each input. The identified symptoms provide valuable insights that could support timely diagnosis and treatment decisions for skin diseases.

Artificial intelligence (AI) is reshaping radiology by enhancing diagnostic accuracy, optimizing workflows, and supporting clinical decision-making. Despite over 500 FDA-approved algorithms, adoption remains limited due to ethical, legal, and operational challenges. Key concerns include data privacy, algorithmic bias, explainability, and accountability. Inadequate representation in training datasets can perpetuate healthcare disparities, while “black-box” decision-making undermines trust and complicates liability. Ethical governance must integrate transparency, fairness, and human oversight from system design through implementation. Data protection frameworks, such as GDPR and Türkiye’s KVKK, mandate anonymization, informed consent, and secure handling of imaging data. Privacy safeguards—metadata cleaning, pixel-level masking, and defacing—are essential to prevent re-identification. Commercial use of health data requires explicit consent, strict oversight, and equitable benefit-sharing. Explainable AI techniques and human-in-the-loop designs can improve trust and reliability. In Türkiye, AI-specific regulations are emerging, with the 2024 Draft AI Law introducing risk-based classification and governance requirements. Global frameworks, including FUTURE-AI and UNESCO guidelines, provide structured pathways for ethical integration. Ensuring AI in radiology is safe, fair, and socially responsible depends on embedding ethical principles into every stage of its lifecycle, fostering interdisciplinary collaboration, and maintaining active oversight to align technological progress with human dignity.