Viet-Anh on Software Logo

What is: TSRUp?

SourceTransformation-based Adversarial Video Prediction on Large-Scale Data
Year2000
Data SourceCC BY-SA - https://paperswithcode.com

TSRUp, or Transformation-based Spatial Recurrent Unit p, is a modification of a ConvGRU used in the TriVD-GAN architecture for video generation.

It largely follows TSRUc, but computes θ\theta, uu and cc in parallel given x_tx\_{t} and h_t1h\_{t−1}, yielding the following replacement for the cc update equation:

c=ρ(W_c_n[h_t1;x_t]+b_c)c = \rho\left(W\_{c} \star\_{n}\left[h\_{t-1}; x\_{t}\right] + b\_{c} \right)

In these equations σ\sigma and ρ\rho are the elementwise sigmoid and ReLU functions respectively and the _n\star\_{n} represents a convolution with a kernel of size n×nn \times n. Brackets are used to represent a feature concatenation.