Skip to content

[Patterns 2023] Tracking Online Low-Rank Approximations of Higher-Order Incomplete Streaming Tensors. In Patterns (Cell Press) 2023.

License

Notifications You must be signed in to change notification settings

thanhtbt/Tensor_Tracking

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Tensor Tracking: Tracking Online Low-Rank Approximations of Higher-Order Incomplete Streaming Tensors

We propose two new provable algorithms for tracking online low-rank approximations of higher-order streaming tensors in the presence of missing data. The first algorithm, dubbed adaptive CP decomposition (ACP), minimizes an exponentially weighted recursive least-squares cost function to obtain the tensor factors in an efficient way, thanks to the alternative minimization framework and the randomized sketching technique. Under the Tucker model, the second algorithm called adaptive Tucker decomposition (ATD), which is more flexible than the first one, first tracks the underlying low-dimensional subspaces covering the tensor factors, and then estimates the core tensor using a stochastic approximation.

Both algorithms are fast and require low computational complexity and memory storage.

DEMO

  • Run files "demo_ACP_xyz.m" and "demo_ATD_xyz.m" for synthetic experiments.
  • Run files "demo_real_video_tracking_completion.m" for the online tensor completion. Video datasets can be downloaded from Releases.

State-of-the-art algorithms for comparison

Some Experimental Results

  • Streaming CP Decomposition

  • Streaming Tucker Decomposition

  • Video Completion

Reference

If you use this code, please cite the following paper.

[1] L.T. Thanh, K. Abed-Meraim, N. L. Trung and A. Hafiane. “Tracking Online Low-Rank Approximations of Higher-Order Incomplete Streaming Tensors”. CellPress Patterns, 2023, [CellPress], [Techrxiv], [PDF].