Polynomial Matrix Completion for Missing Data Imputation and Transductive Learning

Authors: Jicong Fan, Yuqian Zhang, Madeleine Udell3842-3849

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Comparative studies on synthetic data, subspace clustering with missing data, motion capture data recovery, and transductive learning verify the superiority of our methods over the state-of-the-art. and 7 Experiments We compare the proposed PMC methods with LRMC method (nulcear norm minimization), SRMC (Fan and Chow 2017), LADMC (Ongie et al. 2018) (the iterative algorithm), VMC-2 (2-order polynomial kernel), VMC-3 (3order polynomial kernel)(Ongie et al. 2017), and NLMC
Researcher Affiliation Academia Cornell University, Ithaca, NY 14853, USA {jf577, yz2557, udell}@cornell.edu
Pseudocode Yes Algorithm 1 Optimization for PMC using Adam+ Input: X, Ω, k( , ), s or w, γ = 10 6, λ = 10 4, ε = 10 6, tmax, β1 = 0.9, β2 = 0.999, ϵ = 10 8, t = 0 1: initialize [ ˆ X] Ω = 0, ˆx = vec([ ˆ X] Ω), m0 = υ0 = 0 2: repeat 3: t t + 1 4: perform SVD: K( ˆ X) = V SV T 5: compute Θt 1 and ˆ XL( ˆ X, Θt 1) 6: gt vec [ ˆ XL( ˆ X, Θt 1)] Ω 7: mt β1mt 1 + (1 β1)gt 8: υt β2υt 1 + (1 β2)g2 t 9: ˆmt m/(1 βt 1); ˆυt υ/(1 βt 2) 10: ˆxt ˆxt 1 λ ˆmt/( ˆυt + ϵ); [ ˆ X] Ω ˆxt 11: if Lt > Lt 1 12: λ 0.8λ else λ 1.1λ 13: endif 14: until |ˆxt ˆxt 1| < ε or t = tmax Output: ˆ X
Open Source Code No No explicit statement or link providing access to the source code for the methodology was found.
Open Datasets Yes Similar to (Yang, Robinson, and Vidal 2015; Ongie et al. 2017), we perform subspace clustering with missing data on the Hopkins 155 dataset (Tron and Vidal 2007). and We use the trials #1 and #6 of subject #56 of the CMU Mocap dataset.
Dataset Splits No The proportion of training (labeled) data is 50%. No explicit description of a dedicated validation set or a complete train/validation/test split for all experiments was found.
Hardware Specification No No specific hardware details (e.g., GPU/CPU models, memory, or specific computing environments) used for running experiments were provided.
Software Dependencies No The paper mentions using the 'Adam algorithm' but does not specify any software libraries, frameworks, or their version numbers used for implementation (e.g., Python, PyTorch, TensorFlow versions).
Experiment Setup Yes Algorithm 1 Optimization for PMC using Adam+ Input: X, Ω, k( , ), s or w, γ = 10 6, λ = 10 4, ε = 10 6, tmax, β1 = 0.9, β2 = 0.999, ϵ = 10 8, t = 0