Improved Bounded Matrix Completion for Large-Scale Recommender Systems

Authors: Huang Fang, Zhang Zhen, Yiqun Shao, Cho-Jui Hsieh

IJCAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results on real world datasets show that our algorithm can reach a lower objective function value, obtain a higher prediction accuracy and have better scalability compared with existing bounded matrix completion approaches.
Researcher Affiliation Academia Huang Fang1, Zhen Zhang1, Yiqun Shao2, Cho-Jui Hsieh1 1Departments of Statistics and Computer Science 2Department of Mathematics University of California, Davis {hgfang, ezzhang, yqshao, chohsieh}@ucdavis.edu
Pseudocode Yes Algorithm 1 BMC-ADMM for Bounded Matrix Completion
Open Source Code No The paper does not provide any statement or link indicating that the source code for the methodology is openly available.
Open Datasets Yes We use five real world datasets to test the performance of the above matrix completion algorithms, and the detailed data statistics are listed in Table 1. ... dataset m n |Ω| range movielens100k 671 9,066 100,004 [0.5, 5] movielens10m 71,567 10,677 10,000,054 [0.5, 5] Flixster subset 14,761 4,903 81,495 [0.5, 5] Flixster 147,612 48,794 8,196,077 [0.5, 5] Jester 50,692 150 1,728,847 [-10, 10]
Dataset Splits No We randomly split 80% as training data and 20% as testing data. For each algorithm and dataset, we choose the best regularization parameters from λ {0, 0.01, 0.1, 1, 10, 100} based on validation set. However, the paper does not specify the exact split percentage or size of the validation set.
Hardware Specification Yes All the following experiments are conducted on an Intel i5-4590 3.30 GHz CPU with 16GB RAM.
Software Dependencies No The paper mentions using PROPACK for SVD computation but does not provide specific version numbers for any software dependencies.
Experiment Setup Yes We set ρ1 = ρ2 = 1 and try λ = 0.1 and 10, and compare the performance on all the datasets. For each algorithm and dataset, we choose the best regularization parameters from λ {0, 0.01, 0.1, 1, 10, 100} based on validation set. To have a fair comparison, we tried different settings of rank k = 5, 10, 30 for all the algorithms.