A Population-to-individual Tuning Framework for Adapting Pretrained LM to On-device User Intent Prediction

TL;DR

This paper introduces PITuning, a novel framework that adapts pre-trained language models for on-device user intent prediction by modeling event transitions and handling long-tailed preferences, improving personalization and prediction accuracy.

Contribution

The paper presents PITuning, a population-to-individual tuning method that enhances pre-trained models for personalized intent prediction on smartphones, addressing diverse event sequences and long-tailed user preferences.

Findings

PITuning outperforms existing methods on real-world datasets.

It effectively captures long-tailed user preferences.

Demonstrates practicality for on-device deployment.

Abstract

Mobile devices, especially smartphones, can support rich functions and have developed into indispensable tools in daily life. With the rise of generative AI services, smartphones can potentially transform into personalized assistants, anticipating user needs and scheduling services accordingly. Predicting user intents on smartphones, and reflecting anticipated activities based on past interactions and context, remains a pivotal step towards this vision. Existing research predominantly focuses on specific domains, neglecting the challenge of modeling diverse event sequences across dynamic contexts. Leveraging pre-trained language models (PLMs) offers a promising avenue, yet adapting PLMs to on-device user intent prediction presents significant challenges. To address these challenges, we propose PITuning, a Population-to-Individual Tuning framework. PITuning enhances common pattern extraction through dynamic event-to-intent transition modeling and addresses long-tailed preferences via adaptive unlearning strategies. Experimental results on real-world datasets demonstrate PITuning's superior intent prediction performance, highlighting its ability to capture long-tailed preferences and its practicality for on-device prediction scenarios.