Online Tensor Max-Norm Regularization via Stochastic Optimization

Authors: Tong Wu

TMLR 2024 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Numerical experiments demonstrate encouraging results for the effectiveness and robustness of our algorithm. The code is available at https://github.com/twugithub/2024-TMLR-OMRTD. ... In this section, we present several experimental results on both synthetic and real data.
Researcher Affiliation Industry Tong Wu EMAIL Beijing Institute for General Artificial Intelligence
Pseudocode Yes Algorithm 1 t-SVD for third-order tensors ... Algorithm 2 Online Max-Norm Regularized Tensor Decomposition ... Algorithm 3 Updating tensor columns of M, R and E ... Algorithm 4 Data Projection (Problem (10)) ... Algorithm 5 Bisection Method for Solving Problem (12) ... Algorithm 6 The Update of L
Open Source Code Yes The code is available at https://github.com/twugithub/2024-TMLR-OMRTD.
Open Datasets Yes on the CAMO-UOW dataset (Li et al., 2017) for video background subtraction.
Dataset Splits No The dataset contains 10 real video sequences and we use all these sequences for both qualitative and quantitative analysis. No specific train/test/validation splits are mentioned for model training or evaluation on the dataset.
Hardware Specification Yes All experiments are conducted on a PC with an AMD Ryzen 9 5950X 3.40GHz CPU and 64GB RAM with Matlab R2023b.
Software Dependencies Yes All experiments are conducted on a PC with an AMD Ryzen 9 5950X 3.40GHz CPU and 64GB RAM with Matlab R2023b.
Experiment Setup Yes We set λ1 = λ2 = 1/ n1 for OMRTD/r OMRTD, and we follow the default parameter settings for the baselines. ... (we set ϵ = 0.01 in our experiments). ... Initialize: R(0) = E (0) = J (0) = 0, γ = 1.9, µ(0) = 0.1, µmax = 1010, ε = 10 6, and ζ = 0.