Call for Papers
Special Issue on AI, LLMs & Graph Theory: Reimagining Financial Analytics
This special issue delves into the transformative potential of graph theory in the realms of artificial intelligence, finance, and analytics. By exploring innovative applications and methodologies, we aim to uncover how graph-based approaches can enhance predictive modeling, risk assessment, and decision-making processes.
Contributions that highlight cutting-edge research, case studies, and practical implementations that demonstrate the power of graph theory to revolutionize these fields are welcomed!
Introduction:
The intersection between AI and economics presents a unique and important area for research, as both fields share commonalities that can be explored to advance knowledge, understanding, and impact. One of the unique aspects of this intersection is the role of stochasticity, which is present in almost all problems in economics and finance. Neural networks (NN) and machine learning (ML) problems typically can only learn from "a single path" of stochastic outcomes in the absence of a stochastic generator, which presents challenges in developing accurate models and algorithms that can effectively capture stochasticity. Another issue is the oversimplification of economic systems through parsimonious systems of equations commonly seen today, which can lead to inaccuracies in modeling and predicting policy outcomes. Additionally, there are challenges related to the exponential growth of control variables, which can quickly become overwhelming even for the largest hardware simulators, as well as addressing moral hazard issues related to economics and finance, such as global pricing or output constraints, effective policing of socioeconomic AI systems, and algorithms to prevent misuse and abuse. To address these challenges, there is a need for research that explores the intersection between computer science and economics to develop accurate models and algorithms that can effectively capture the stochastic nature of economic systems, increase the explainability of fitted systems, address the challenges of exponential growth in control variables, and navigate the moral hazard issues algorithmically. Researchers are invited to submit original and unpublished work that advances our understanding of the interplay between AI and graph learning and their effectiveness in solving real-world problems in economics and finance via computational models.
Topics covered:
Topics of interest include but are not limited to:
- Theoretical and experimental studies on computational frameworks for graph theory and Lagrangian Neural Networks
- Implied graphs and similar methodology to describe the underlying structure of a system of multiple time series data, including novel methods to describe the underlying structure of the economic/financial systems and analyze the relationships between different variables
- Spectral techniques and similar methods to reduce the dimensionality of large-scale graphs used to analyze real-world economic/financial AI models
- Hybridization of semi-analytical methods and heuristics to encourage faster convergence of solving real-world problems in economics/finance
- Including uncertainty and stochasticity in large-scale AI and graph learning
- Multi-objective algorithms and their integration with global economic/financial constraints
- Impact of estimators and signal processing on key variables as graph data inputs to global economic/financial problems
- Graphs for financial Big Data analytics, such as in large-scale portfolio optimization and “Limit Order Book modelling"
- Computational benchmark performance on representative problems on the latest hardware such as GPUs and Quantum Computers, to better understand the limitations of what real-world problems can be solved effectively today
- An alternative approach to extreme value theory for machine learning models, or a method that can provide sensible out-of-sample values with economic significance, such as in the context of climate impact estimates beyond 2 degrees
- ML under distribution/covariance regime shifts, when the actual predictor to outcome relationship changes, for example, how marketing models done on one date can be different from the acted-on policy on a different date, also known as concept shift.
- Xiaoli Li, Institute for Infocomm Research, and Nanyang Technological University, Singapore
https://www.a-star.edu.sg/cfar/about-cfar/our-team/dr-li-xiaoli
- Anthony G. Constantinides, Imperial College London, UK, a.constantinides@imperial.ac.uk
- Bernard Lee, HedgeSPA, USA, bernard.lee@hedgespa.com or bernardl@cs.stanford.edu
- Danilo Mandic, Imperial College London, UK, d.mandic@imperial.ac.uk
- Submission deadline: 1 Sept 2025
- First-round review decisions: 15 Oct 2025
- Deadline for revision submissions: 30 Nov 2025
- Notification of final decisions: 31 Dec 2025
- Tentative publication: 15 Feb 2026
You are invited to submit your articles no later than the given deadline. All papers will be subject to a rigorous peer review. Conference papers with at least 30% new content will also be considered (If copyrighted material is used, the author should obtain the necessary copyright release and submit it along with the manuscript)
- Please read the [Guide for Authors] before submitting and prepare your manuscript using the journal template of WSARAI: Latex2e : (readme / Download) MS-Word: (ZIP files) (readme / Download)
- All articles should be [submitted online], please select [Special Issue on AI, LLMs & Graph Theory: Reimagining Financial Analytics] on submission.