LATTE: Learning Aligned Transactions and Textual Embeddings for Bank Clients

TL;DR

LATTE is a contrastive learning framework that efficiently aligns client event embeddings with semantic embeddings from frozen LLMs, improving sequence representation while reducing computational costs in financial applications.

Contribution

LATTE introduces a novel contrastive learning approach that aligns raw event embeddings with frozen LLM semantic embeddings, enabling efficient and effective client behavior modeling.

Findings

Outperforms state-of-the-art sequence representation methods

Reduces inference cost and input size significantly

Effective in latency-sensitive financial environments

Abstract

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.