Learning Better Intent Representations for Financial Open Intent Classification
This work addresses the problem of handling unseen intents for banks and financial entities using virtual agents, representing an incremental improvement over existing adaptive decision boundary methods.
The paper tackles open intent classification in financial virtual agents by improving intent representations through supervised pre-training methods like prefix-tuning and fine-tuning the last layer of a large language model, achieving a 1.63% to 2.07% higher accuracy than prior state-of-the-art methods on benchmarks like banking77 with only 0.1% additional trainable parameters.
With the recent surge of NLP technologies in the financial domain, banks and other financial entities have adopted virtual agents (VA) to assist customers. A challenging problem for VAs in this domain is determining a user's reason or intent for contacting the VA, especially when the intent was unseen or open during the VA's training. One method for handling open intents is adaptive decision boundary (ADB) post-processing, which learns tight decision boundaries from intent representations to separate known and open intents. We propose incorporating two methods for supervised pre-training of intent representations: prefix-tuning and fine-tuning just the last layer of a large language model (LLM). With this proposal, our accuracy is 1.63% - 2.07% higher than the prior state-of-the-art ADB method for open intent classification on the banking77 benchmark amongst others. Notably, we only supplement the original ADB model with 0.1% additional trainable parameters. Ablation studies also determine that our method yields better results than full fine-tuning the entire model. We hypothesize that our findings could stimulate a new optimal method of downstream tuning that combines parameter efficient tuning modules with fine-tuning a subset of the base model's layers.