Greedy Learning of Generalized Low-Rank Models

Authors: Quanming Yao, James T. Kwok

IJCAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results show that it is much faster than the state-of-the-art, with comparable or even better prediction performance.In this section, we compare the proposed algorithms with the state-of-the-art on link prediction and robust matrix factorization. Experiments are performed on a PC with Intel i7 CPU and 32GB RAM. All the codes are in Matlab.
Researcher Affiliation Academia Quanming Yao James T. Kwok Department of Computer Science and Engineering Hong Kong University of Science and Technology Hong Kong {qyaoaa, jamesk}@cse.ust.hk
Pseudocode Yes Algorithm 1 R1MP [Wang et al., 2014].Algorithm 2 GLRL for low-rank matrix learning with smooth convex objective f.Algorithm 3 GLRL for low-rank matrix learning with nonsmooth objective f.
Open Source Code No The paper does not provide an explicit statement or link for the open-source code of their proposed GLRL/EGLRL method. The only GitHub link provided is for a baseline method (AIS-Impute).
Open Datasets Yes Experiments are performed on the Epinions and Slashdot data sets2 [Chiang et al., 2014] (Table 1)... 2https://snap.stanford.edu/data/Experiments are performed on the Movie Lens data sets4 (Table 3)... 4http://grouplens.org/datasets/movielens/
Dataset Splits Yes Following [Chiang et al., 2014], we use 10-fold cross-validation and fix the rank r to 40.50% of the ratings are randomly sampled for training while the rest for testing.
Hardware Specification Yes Experiments are performed on a PC with Intel i7 CPU and 32GB RAM.
Software Dependencies No All the codes are in Matlab. The paper does not provide specific version numbers for Matlab or any key libraries/solvers used.
Experiment Setup Yes As in [Wang et al., 2014], we fix the number of power method iterations to 30. Following [Chiang et al., 2014], we use 10-fold cross-validation and fix the rank r to 40.For AIS-Impute and Alt Min, they are stopped when the relative change in the objective is smaller than 10 4.we compare GLRL in Algorithm 3 (with = 0.99 and c2 = 0.05).The ranks used for the 100K, 1M, 10M data sets are 10, 10, and 20, respectively.Experiments are repeated five times with random training/testing splits.