Jump to content

Draft:Large Transaction Model (LTM)

From Wikipedia, the free encyclopedia
  • Comment: You didn't rewrite the draft without LLM usage as the previous reviewer told you to do. 🍅 fx (talk) 20:53, 6 September 2026 (UTC)
  • Comment: Needs to be rewritten from scratch without LLMs. Helpful Raccoon (talk) 00:30, 9 June 2026 (UTC)


A Large Transaction Model (LTM) is a proposed class of machine learning models designed to analyze sequences of financial transactions in order to identify patterns, anomalies, and behavioral trends. The term has been used primarily in industry contexts to describe models trained on large-scale transactional datasets, particularly for applications in fraud detection, anti-money laundering (AML), and financial risk assessment.[1][2][3]

History

[edit]

Early approaches to transaction analysis

[edit]

Financial transaction analysis has long been integral to banking, payment processing, and compliance regulation. Early systems, particularly in the late 1980s, predominantly employed rule-based methods.[4]

As digital financial systems expanded and transaction volumes increased, statistical models and conventional machine learning techniques were introduced to improve detection capabilities.[4]

Introduction of sequence modeling

[edit]

In the 2010s, the availability of financial transactions as successive time series (longitudinal data) led to a shift toward analyzing financial activity as sequences of events instead of isolated transactions.[5] Sequence-based neural networks, including recurrent neural networks (RNNs) and long short-term memory (LSTM) models, were used to process transaction data streams.[5]

Development

[edit]

Adoption of transformer architectures

[edit]

Transformer architectures, first introduced in the field of natural language processing (NLP), represent a major breakthrough in sequence modeling. Models such as BERT and GPT demonstrated that attention-based mechanisms can effectively learn relationships between distant elements in long sequences without relying on recurrence.

Emergence and use of the term

[edit]

The term “Large Transaction Model” emerged in the early 2020s within industry discussions, reflecting an analogy with large language models and emphasizing the scale of both data and model architecture. In October 2023, the Cambridge-based company Featurespace introduced a model called TallierLTM,[1][2][6] describing it as a transformer-based system trained on extensive transaction data for applications such as fraud detection and financial crime prevention.[2]

The use of the term has remained relatively limited and is not yet widely established in academic research.[7][8]

Current research and practical matters

[edit]

Development in this area is affected by both technical and regulatory factors. Research focuses on refining model efficiency, managing sparse and heterogeneous transaction data, and integrating multiple data sources, including graph-based representations of financial relationships.[4]

Application

[edit]

Large Transaction Models are proposed for various financial applications. In fraud detection, they identify transactions that differ from established behavioral patterns. For anti-money laundering, they help detect complex or coordinated financial activity that may indicate illicit behavior.[9] They have also been applied in credit assessment, where analysis of transaction histories is used to support credit adjudication decisions.[10]

References

[edit]
  1. 1 2 Penn, David (2023-10-24). "Featurespace Launches GenAI-Powered Financial Crime Fighting Model, TallierLTM". Finovate. Retrieved 2026-05-22.
  2. 1 2 3 PYMNTS (2023-10-24). "Featurespace Unveils AI 'Large Transaction Model' to Combat Fraud". PYMNTS.com. Retrieved 2026-05-22.
  3. ↑ Mikhailov, Serhii (2024-12-04). "Banks Improve Their AML and KYC Practices". PaySpace Magazine. Retrieved 2026-09-02.
  4. 1 2 3 Valero, César Soto (2025-04-03). "From Classical ML to DNNs and GNNs for Real-Time Financial Fraud Detection". César Soto Valero. Retrieved 2026-05-22.
  5. 1 2 Lucas, Yvan; Jurgovsky, Johannes (2020-10-13). "Credit card fraud detection using machine learning: A survey". arXiv:2010.06479 [cs.LG].
  6. ↑ "Featurespace launches TallierLTM". thepaypers.com. Retrieved 2026-05-22.
  7. ↑ "ATOM Foundation Model". RBC Borealis. Retrieved 2026-05-22.
  8. ↑ Liu, Meng (2025-11-17). "Key Takeaways From The Singapore FinTech Festival's 10th Anniversary". Forrester. Retrieved 2026-05-22.
  9. ↑ Nguyen, Dieu Anh (2025-08-26). "AI in AML: Top Use Cases You Need To Know". SmartDev. Retrieved 2026-09-02.
  10. ↑ "ATOM Foundation Model". RBC Borealis. Retrieved 2026-09-02.