LATTE: Learning Aligned Transactions and Textual Embeddings for Bank Clients
This work addresses computational inefficiencies in using LLMs for financial client embedding tasks, offering a deployable solution for latency-sensitive applications, though it is incremental as it builds on existing contrastive learning and LLM techniques.
The paper tackles the problem of learning client embeddings from long sequences of historical communications in financial applications by proposing LATTE, a contrastive learning framework that aligns raw event embeddings with semantic embeddings from frozen LLMs, significantly reducing inference cost and input size. The method outperforms state-of-the-art techniques on real-world financial datasets while remaining deployable in latency-sensitive environments.
Learning clients embeddings from sequences of their historic communications is central to financial applications. While large language models (LLMs) offer general world knowledge, their direct use on long event sequences is computationally expensive and impractical in real-world pipelines. In this paper, we propose LATTE, a contrastive learning framework that aligns raw event embeddings with semantic embeddings from frozen LLMs. Behavioral features are summarized into short prompts, embedded by the LLM, and used as supervision via contrastive loss. The proposed approach significantly reduces inference cost and input size compared to conventional processing of complete sequence by LLM. We experimentally show that our method outperforms state-of-the-art techniques for learning event sequence representations on real-world financial datasets while remaining deployable in latency-sensitive environments.