Multiple Robust Learning for Recommendation

Authors: Haoxuan Li, Quanyu Dai, Yuru Li, Yan Lyu, Zhenhua Dong, Xiao-Hua Zhou, Peng Wu

AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We conduct extensive experiments on real-world and semi-synthetic datasets, which demonstrates the superiority of the proposed approach over the state-of-the-art methods.
Researcher Affiliation Collaboration 1Peking University 2Huawei Noah s Ark Lab 3Beijing Technology and Business University
Pseudocode Yes Algorithm 1: Alternating Multiple Robust Learning with Stabilization
Open Source Code Yes The proposed MR2 and most existing debiasing methods are model-agnostic, which can be integrated into existing recommendation models for unbiased learning based on biased data. 2https://gitee.com/mindspore/models/tree/master/research/ recommend/multi_robust
Open Datasets Yes We conduct experiments on both real-world datasets and semi-synthetic datasets to evaluate the effectiveness of our proposed method. Datasets. We consider two benchmark real-world datasets containing MNAR and MAR ratings, i.e., Coat3 (Schnabel et al. 2016) and Yahoo4 (Marlin and Zemel 2009), as existing work (Schnabel et al. 2016; Wang et al. 2019). ... We conduct experiments on semi-synthetic datasets constructed from Movie Lens 100K5 (ML-100K).
Dataset Splits Yes All experiments are implemented on Py Torch (Paszke et al. 2019) with Adam optimizer (Kingma and Ba 2015), and grid search is used to choose the optimal set of hyper-parameters based on a validation set split from the training set.
Hardware Specification No The paper does not explicitly describe the hardware used for experiments. It mentions support from 'CANN (Compute Architecture for Neural Networks) and Ascend AI Processor' but this is a general statement about the supporting technology, not the specific hardware used for these experiments.
Software Dependencies No All experiments are implemented on Py Torch (Paszke et al. 2019) with Adam optimizer (Kingma and Ba 2015)... No specific version numbers for PyTorch or Adam are provided, only the citation years.
Experiment Setup Yes All experiments are implemented on Py Torch (Paszke et al. 2019) with Adam optimizer (Kingma and Ba 2015), and grid search is used to choose the optimal set of hyper-parameters based on a validation set split from the training set. ... where I is an identity matrix, and λ is a hyper-parameter for stabilization.