A Survey of Explainable Artificial Intelligence (XAI) in Financial Time Series Forecasting

ACM Computing Surveys, Journal Year: 2025, Volume and Issue: unknown

Published: April 18, 2025

Artificial Intelligence (AI) models have reached a very significant level of accuracy. While their superior performance offers considerable benefits, inherent complexity often decreases human trust, which slows application in high-risk decision-making domains, such as finance. The field eXplainable AI (XAI) seeks to bridge this gap, aiming make more understandable. This survey, focusing on published work from 2018 2024, categorizes XAI approaches that predict financial time series. In paper, explainability and interpretability are distinguished, emphasizing the need treat these concepts separately they not applied same way practice. Through clear definitions, rigorous taxonomy approaches, complementary characterization, examples XAI’s finance industry, paper provides comprehensive view current role It can also serve guide for selecting most appropriate approach future applications.

Language: Английский

---

Decision‐Making in M&A Under Market Mispricing: The Role of Deep Learning Models

Managerial and Decision Economics, Journal Year: 2025, Volume and Issue: unknown

Published: April 4, 2025

ABSTRACT In the ever‐evolving landscape of financial markets, mergers and acquisitions (M&A) play a pivotal role in shaping corporate ecosystem. However, presence market mispricing, driven by various factors such as information asymmetry, behavioral biases, external shocks, has been persistent challenge for investors corporations alike. Understanding intricate relationship between stock mispricing M&A is crucial making informed investment decisions fostering resilient environment. This research explores how impacts within fragmented setting, utilizing deep learning methods to uncover complex patterns relationships. By analyzing inefficiencies, study aims provide deeper understanding influences strategies outcomes. Employing quantitative descriptive design, gathered valid data through distributed questionnaires, yielding responses from 130 traders, 115 participants, 99 regulators policymakers. The analysis was conducted using Statistical Package Social Sciences (SPSS). Firstly, it establishes effectiveness algorithms detecting quantifying providing reliable measure its extent. then differential performance outcomes companies engaging during periods prevalent compared those efficient pricing. study's novel contribution lies introduction sentiment models incorporate participants' sentiments, enhancing accuracy detection impact on activity. Finally, this contributes valuable insights into integration techniques leveraging strategic decision‐making context M&A.

Language: Английский

---

ALL diagnosis: can efficiency and transparency coexist? An explainble deep learning approach

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: April 14, 2025

Acute Lymphoblastic Leukemia (ALL) is a life-threatening malignancy characterized by its aggressive progression and detrimental effects on the hematopoietic system. Early accurate diagnosis paramount to optimizing therapeutic interventions improving clinical outcomes. This study introduces novel diagnostic framework that synergizes EfficientNet-B7 architecture with Explainable Artificial Intelligence (XAI) methodologies address challenges in performance, computational efficiency, explainability. The proposed model achieves improved accuracies exceeding 96% Taleqani Hospital dataset 95.50% C-NMC-19 Multi-Cancer datasets. Rigorous evaluation across multiple metrics-including Area Under Curve (AUC), mean Average Precision (mAP), Accuracy, Precision, Recall, F1-score-demonstrates model's robustness establishes superiority over state-of-the-art architectures namely VGG-19, InceptionResNetV2, ResNet50, DenseNet50 AlexNet . Furthermore, significantly reduces overhead, achieving up 40% faster inference times, thereby enhancing applicability. To opacity inherent Deep learning (DL) models, integrates advanced XAI techniques, including Gradient-weighted Class Activation Mapping (Grad-CAM), (CAM), Local Interpretable Model-Agnostic Explanations (LIME), Integrated Gradients (IG), providing transparent explainable insights into predictions. fusion of high precision, explainability positions as transformative tool for ALL diagnosis, bridging gap between cutting-edge AI technologies practical deployment.

Language: Английский

---

Investigating the impact of sentiments on stock market using digital proxies: Current trends, challenges, and future directions

Expert Systems with Applications, Journal Year: 2025, Volume and Issue: 285, P. 127864 - 127864

Published: May 5, 2025

Language: Английский

---