Noise-Tolerant Life-Long Matrix Completion via Adaptive Sampling

Authors: Maria-Florina F. Balcan, Hongyang Zhang

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

Reproducibility Variable Result LLM Response
Research Type Experimental Our proposed algorithms perform well experimentally in both synthetic and real-world datasets. ... 4 Experimental Results Bounded Deterministic Noise: We verify the estimated error of our algorithm in Theorem 1 under bounded deterministic noise. Our synthetic data are generated as follows. ... Sparse Random Noise: We then verify the exact recoverability of our algorithm under sparse random noise. ... Mixture of Subspaces: To test the performance of our algorithm for the mixture of subspaces, we conduct an experiment on the Hopkins 155 dataset.
Researcher Affiliation Academia Maria-Florina Balcan Machine Learning Department Carnegie Mellon University, USA ninamf@cs.cmu.edu Hongyang Zhang Machine Learning Department Carnegie Mellon University, USA hongyanz@cs.cmu.edu
Pseudocode Yes Algorithm 1 Noise-Tolerant Life-Long Matrix Completion under Bounded Deterministic Noise
Open Source Code No The paper does not provide an explicit statement or link for open-source code for the methodology described.
Open Datasets Yes Mixture of Subspaces: To test the performance of our algorithm for the mixture of subspaces, we conduct an experiment on the Hopkins 155 dataset. The Hopkins 155 database is composed of 155 matrices/tasks...
Dataset Splits No The paper does not explicitly state training, validation, and test dataset splits with percentages or sample counts. It mentions "randomly pick 20% entries to be unobserved" but this refers to missing data within samples, not dataset splits.
Hardware Specification No No specific hardware details (e.g., CPU, GPU models, memory) used for running the experiments are provided in the paper.
Software Dependencies No The paper does not provide any specific software dependencies or version numbers (e.g., programming languages, libraries, frameworks) used for the experiments.
Experiment Setup Yes Our synthetic data are generated as follows. We construct 5 base vectors {ui}5 i=1 by sampling their entries from N(0, 1). The underlying matrix L is then generated by L = h u11T 200, P2 i=1 ui1T 200, P3 i=1 ui1T 200, P4 i=1 ui1T 200, P5 i=1 ui1T 1,200 i R100 2,000, each column of which is normalized to the unit ℓ2 norm. Finally, we add bounded yet unstructured noise to each column, with noise level ϵnoise = 0.6. We randomly pick 20% entries to be unobserved. ... The sparse random noise is drawn from standard Gaussian distribution such that s0 d r 1. For each size of problem (50 500 and 100 1, 000), we test with different rank ratios r/m and measurement ratios d/m. The experiment is run by 10 times.