Kai Yin, Xiangjue Dong, Chengkai Liu, Allen Lin, Lingfeng Shi, Ali Mostafavi, James Caverlee. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Deep neural networks have achieved great success in sequential recommendation systems. While maintaining high competence in user modeling and next-item recommendation, these models have long been plagued by the numerous parameters and computation, which inhibit them to be deployed on resource-constrained mobile devices. Model quantization, as one of the main paradigms for compression techniques, converts float parameters to low-bit values to reduce parameter redundancy and accelerate inference. To avoid drastic performance degradation, it usually requests a fine-tuning phase with an original dataset. However, the training set of user-item interactions is not always available due to transmission limits or privacy concerns. In this paper, we propose a novel framework to quantize sequential recommenders without access to any real private data. A generator is employed in the framework to synthesize fake sequence samples to feed the quantized sequential recommendation model and minimize the gap with a full-precision sequential recommendation model. The generator and the quantized model are optimized with a min-max game — alternating discrepancy estimation and knowledge transfer. Moreover, we devise a two-level discrepancy modeling strategy to transfer information between the quantized model and the full-precision model. The extensive experiments of various recommendation networks on three public datasets demonstrate the effectiveness of the proposed framework.

Sequential recommendation, which aims to leverage users' historical behaviors to predict their next interaction, has become a research hotspot in the field of recommendation. Time is one of the important contextual information for interaction. However, most previous works only use time information as a model feature or time prediction as an auxiliary task and ignore the duality between sequential recommendation task and time prediction task. Compared with the method of sharing parameters in multi-task learning, this paper proposed a dual learning framework to jointly model two tasks and incorporate the probabilistic dual properties between them in the training stage. In addition, we design an appropriate base model for each task. Finally, experiments on two public datasets demonstrated the effectiveness of the proposed dual learning framework in sequential recommendation scenarios.