arXiv subjects

Kiarash Shamsi

Publications and source records attributed to Kiarash Shamsi.

5 recordsLinked to original sources

arXiv2026

Edge Sparsification via Temporal Forman-Ricci Curvature for Dynamic Graph Learning

Temporal graph learning has become essential for analyzing real-world systems whose interactions continuously evolve over time, including financial transaction networks, communication systems, and online social platforms. However, learning from large-scale temporal graphs remains computationally challenging when networks are dense and rapidly changing. To address this limitation, we propose a network-curvature-inspired edge sparsification framework for dynamic graph learning. Our proposed method, TRicci, extends classical Forman-Ricci curvature to directed weighted temporal graphs by capturing structural support, temporal recency, and local interaction competition. Experiments on 9 transaction networks and 3 temporal graph benchmark datasets demonstrate that the proposed framework preserves predictive performance across multiple graph-level prediction tasks. The results show that TRicci sparsifies temporal graphs by approximately 80% while reducing end-to-end downstream training and inference time by an average of 55.94%, without substantial degradation in predictive performance. Our findings suggest that temporal curvature can serve as a principled basis for scalable temporal graph learning by preserving predictive temporal-structural information under substantial sparsification.

cs.LG↗

arXiv2026

LendNova: Towards Automated Credit Risk Assessment with Language Models

Credit risk assessment is essential in the financial sector, but has traditionally depended on costly feature-based models that often fail to utilize all available information in raw credit records. This paper introduces LendNova, the first practical automated end-to-end pipeline for credit risk assessment, designed to utilize all available information in raw credit records by leveraging advanced NLP techniques and language models. LendNova transforms risk modeling by operating directly on raw, jargon-heavy credit bureau text using a language model that learns task-relevant representations without manual feature engineering. By automatically capturing patterns and risk signals embedded in the text, it replaces manual preprocessing steps, reducing costs and improving scalability. Evaluation on real-world data further demonstrates its strong potential in accurate and efficient risk assessment. LendNova establishes a baseline for intelligent credit risk agents, demonstrating the feasibility of language models in this domain. It lays the groundwork for future research toward foundation systems that enable more accurate, adaptable, and automated financial decision-making.

cs.LG↗

arXiv2025

MiNT: Multi-Network Training for Transfer Learning on Temporal Graphs

Temporal Graph Learning (TGL) has become a robust framework for discovering patterns in dynamic networks and predicting future interactions. While existing research has largely concentrated on learning from individual networks, this study explores the potential of learning from multiple temporal networks and its ability to transfer to unobserved networks. To achieve this, we introduce Temporal Multi-network Training MiNT, a novel pre-training approach that learns from multiple temporal networks. With a novel collection of 84 temporal transaction networks, we pre-train TGL models on up to 64 networks and assess their transferability to 20 unseen networks. Remarkably, MiNT achieves state-of-the-art results in zero-shot inference, surpassing models individually trained on each network. Our findings further demonstrate that increasing the number of pre-training networks significantly improves transfer performance. This work lays the groundwork for developing Temporal Graph Foundation Models, highlighting the significant potential of multi-network pre-training in TGL.

cs.LG↗

arXiv2021

A Fair Method for Distributing Collective Assets in the Stellar Blockchain Financial Network

The financial industry is a pioneer in Blockchain technology. One of the most popular platforms in Token-based banking is the flexible Stellar platform. This platform is open-source, and today, its wide range of features makes it possible for many countries and companies to use it in cryptocurrency and Token-based modern banking. This network charges a fee for each transaction. As well, a percentage of the net amount is generated as the inflation rate of the network due to the increased number of tokens. These fees and inflationary amounts are aggregated into a general account and ultimately distributed among members of the network on a collective vote basis. In this mechanism, network users select an account as the destination for which they wish to transfer assets using their user interface, which is generally a wallet. This account could be the account of charities that need this help. It is then determined the target distribution network based on the voting results of all members. One of the challenges in this network is the purposeful and fair distribution of these funds between accounts. In this paper, the first step is a complete infrastructure of a Stellar financial network that will consist of three network-based segments of the core network, off-chain server, and wallet interface. In the second step, a context-aware recommendation system will be explored and implemented as a solution for the purposeful management of payroll account selection. The results of this study concerning the importance of the purposeful division of collective assets and showing a context-aware recommendation system as a solution to improve the process of stellar users' participation in the voting process by effectively helping them in choosing an eligible destination

cs.DC↗

arXiv2020

A Secure and Efficient Approach for Issuing KYC Token As COVID-19 Health Certificate Based on Stellar Blockchain Network

Today's world is struggling with the COVID-19 pandemic, as one of the greatest challenges of the 21st century. During the lockdown caused by this disease, many financial losses have been inflicted on people and all industries. One of the fastest ways to save these industries from the COVID-19 is to provide a reliable solution for people's health assessment. In this article, blockchain technology is used to propose a model which provides and validates the health certificates for people who travel or present in society. For this purpose, we take advantage of blockchain features in protecting people's privacy. Since a variety of antibody and human health proving tests against the virus are developing, this study tries simultaneously to design an integrated and secure system to meet the authenticity and accuracy of different people's health certificates for the companies requiring these certifications. In this system, on the one hand, there are qualified laboratories that are responsible for performing standard testing and also providing results to the system controller. Finally, people are considered as the end-user of the system. To provide test information for the entities, the mechanism of KYC tokens will be used based on the Stellar private blockchain network. In this mechanism, the user will buy a certain amount of KYC tokens from the system controller. These tokens are charged in the user's wallet, and the user can send these tokens from his wallet to any destination company, to exchange the encrypted health certificate information. Finally, considering the appropriate platform provided by blockchain technology and the requirement of a reliable and accurate solution for issuing health certificates during the Covid-19 pandemic or any other disease, this article offers a solution to meet the requirements.

cs.CR↗