Paper: arXiv 2309.11960
Authors: Wei Jie Yeo, Wihan van der Heever, Rui Mao, Erik Cambria, Ranjan Satapathy, Gianmarco Mengaldo
Abstract
The success of artificial intelligence (AI), and deep learning models in particular, has led to their widespread adoption across various industries due to their ability to process huge amounts of data and learn complex patterns. However, due to their lack of explainability, there are significant concerns regarding their use in critical sectors, such as finance and healthcare, where decision-making transparency is of paramount importance. In this paper, we provide a comparative survey of methods that aim to improve the explainability of deep learning models within the context of finance. We categorize the collection of explainable AI methods according to their corresponding characteristics, and we review the concerns and challenges of adopting explainable AI methods, together with future directions we deemed appropriate and important.
Complexity vs Empirical Score
- Math Complexity: 4.0/10
- Empirical Rigor: 3.0/10
- Quadrant: Philosophers — conceptual discussion, lower math and empirical scores
Why this score: The paper is a review/survey discussing conceptual frameworks and categorizations of explainable AI methods, with no original mathematical derivations or empirical testing presented. Its focus is on literature synthesis rather than algorithmic innovation or backtesting.
Research Flowchart
flowchart TD A["Research Goal: Survey Explainable AI<br>in Financial Deep Learning"] --> B["Methodology: Comparative Literature Review"] B --> C["Data/Inputs: Financial XAI Methods<br>from Academic Literature"] C --> D["Computational Process: Categorization &<br>Analysis by Method Characteristics"] D --> E["Key Outcomes: Comparative Survey,<br>Challenges, and Future Directions"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper provides an extensive review of XAI methods for the financial domain (FinXAI) and proposes a sequential decision‑making framework for selecting appropriate FinXAI tools, emphasizing audience‑centric selection.
Source sentence
In this paper, we provide an extensive review of XAI methods for the financial field that we name FinXAI. The main contributions of our work as such: • We provide an extensive study on consolidating XAI methods in the field of finance (FinXAI), for researchers interested in prioritizing transparency in their solutions. • We frame the FinXAI process as a sequential flow of decision-making processes (see Fig 4), where
Assumes 1
- AI awareness is achieved, i.e., it is known that AI is involved in the financial process. model
Source sentence
In this review, we assume that AI awareness is achieved, i.e., we know that in a given financial process AI is involved, and focus on AI explainability, also referred to as eXplainable AI or simply XAI.
Data
No dataset named in the text.
Costs
Not applicable
Finds 5
- The majority of explanation types in the literature are limited to fact‑based explanations, leading the authors to explicitly search for counterfactual techniques. other
Source sentence
We also noticed the majority of explanation types were limited to fact-based explanations, hence we explicitly search for techniques explaining in the form of counterfactuals.
- Both inherently transparent models and post‑hoc explainability techniques are covered, with a recommendation to further improve inherently transparent models. other
Source sentence
This includes both inherently transparent models and post-hoc explainability techniques, the former of which we advocate for more improvements to be made.
- Key challenges identified include over‑reliance on potentially misleading explanations and the limited incorporation of multimodal (especially textual) information in XAI methods. other
Source sentence
8.1 Over-reliance A means of interpreting the model can be helpful while transforming how users interact with data. 8.7 Multimodal XAI A less discussed avenue for improvement is in the incorporation of multimodal information, particularly natural language.
- The review categorizes financial XAI works into three task areas—credit evaluation, financial prediction, and financial analytics—and provides detailed classification tables. other
Source sentence
stock index prediction, market condition forecasting, volatility forecasting, algorithmic trading, financial growth rate, economic crisis forecast, bankruptcy prediction, fraud detection, mortgage default) and financial analytics (financial text classification, spending behavior, financial corporate social responsibility J. ACM, Vol. 37, No. 4, Article 111. Following the task classification, we further differentiate
- An audience‑centric XAI approach is highlighted as essential, emphasizing the need to align explanation techniques with the intended users. other
Source sentence
Hence, we highlight the importance of an audience-centric XAI technique as a more realistic expectation.
Checked by others
No other paper in the archive has been found to support or contradict these findings. Coverage is partial, so this is neither replication nor refutation.
1 paper in the archive
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.