Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1]
Triple Component Matrix Factorization: Untangling Global, Local, and Noisy Components
Authors: Naichen Shi, Salar Fattahi, Raed Al Kontar
JMLR 2024 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In this section, we investigate the numerical performance of TCMF on several datasets. We first use synthetic datasets to verify the convergence in Theorem 5 and validate TCMF’s capability in recovering the common and individual features from noisy observations. Then, we use two examples of noisy video segmentation and anomaly detection to illustrate the utility of common, unique, and noise components. |
| Researcher Affiliation | Academia | Naichen Shi EMAIL Department of Industrial & Operations Engineering University of Michigan Ann Arbor, MI 48109, USA Salar Fattahi EMAIL Department of Industrial & Operations Engineering University of Michigan Ann Arbor, MI 48109, USA Raed Al Kontar EMAIL Department of Industrial & Operations Engineering University of Michigan Ann Arbor, MI 48109, USA |
| Pseudocode | Yes | Algorithm 1 TCMF: alternating minimization Algorithm 2 JIMF by heterogeneous matrix factorization Algorithm 3 JIMF by personalized PCA |
| Open Source Code | Yes | Code is available in the linked Github repository. |
| Open Datasets | Yes | We use a surveillance video from Vacavant et al. (2013) as an example. In this study, the dataset (Jin et al., 2000; Yan et al., 2018) comes from the Hot Eye video of a rolling steel plate. |
| Dataset Splits | Yes | In the case study, the threshold is set to be the highest value in the first 50 frames, which is the in-control group that does not contain anomalies (Yan et al., 2018). |
| Hardware Specification | Yes | Experiments in this section are performed on a desktop with 11th Gen Intel(R) i7-11700KF and NVIDIA Ge Force RTX 3080. |
| Software Dependencies | No | We implement Algorithm 1 with HMF (Shi et al., 2023) as its subroutine JIMF. ... The subroutine JIMF in Algorithm 1 is implemented by HMF with spectral initialization. |
| Experiment Setup | Yes | With the generated {M(i)}, we run Algorithm 1 with ρ = 0.99 to estimate local, global, and sparse components. The subroutine JIMF in Algorithm 1 is implemented by HMF with spectral initialization. ... for each call of HMF, we run 500 iterations with constant stepsize 0.005. For the video segmentation task, we set the stepsize to 5e-6 and R = 200. And on the hot rolling data, we set the stepsize to 4e-5 and R = 500. In our experiments, we observe that the regularization parameter β exerts a negligible influence on the convergence of Algorithm 2. Consequently, we maintain β within the range of 10e-6 to 10e-5 in all our experiments. |