TGANv2: Efficient Training of Large Models for Video Generation with Multiple Subsampling Layers
By Masaki Saito, Shunta Saito. 2018
46 citations recorded by the source corpus.
Subfields: Computer Vision and Pattern Recognition, Generative Models
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Jitendra Malik
- Dacheng Tao
- Ming-Hsuan Yang
- Trevor Darrell
- Andrew Zisserman
- Abhinav Kumar Gupta
- Xiaogang Wang
In Generative Models
- Karen Simonyan
- Ali Razavi
- Aäron van den Oord
- Jacob Walker
- Jeff Donahue
- Aidan Clark
- Jonathan Ho
- Stefano Ermon
Citations in the corpus
- VideoGPT: Video Generation using VQ-VAE and Transformers (2021, neutral): “In addition, (Saito & Saito, 2018) and (Kahembwe & Ramamoor-thy, 2020) propose more scalable and efficient GAN models for training on less compute.”
- High Fidelity Speech Synthesis with Adversarial Networks (2019, neutral): “Similar approaches have also been used in image-to-image transfer (Huang et al., 2018) and video synthesis (Saito & Saito, 2018).”
- Predicting Video with VQVAE (2021, background): “While some work in the area of video generation [7, 54, 45] has explored video synthesis—generating videos with no prior frame information—many approaches actually focus on the task of video prediction conditioned on past frames [41, 47, 38, 30, 23, 2, 34, 58, 60, 13, 27].”
- Adversarial Video Generation on Complex Datasets (2019, methodology): “TGANv2 (Saito & Saito, 2018) proposes “adaptive batch reduction” for efficient training, an operation which randomly samples subsets of videos within a batch and temporal subwindows within each video.”
- Latent Video Transformer (2020, background): “Later GANs were applied to address those issues, but they suffer from mode-dropping behavior.”
- Lower Dimensional Kernels for Video Discriminators (2019, methodology): “Saito nd Saito (2018) explore sub-sampling as a way to scale video AN models to 192 × 192 video generation and achieve state-ofhe-art results when combined with a large batch training regime. heir proposed model, TGANv2,3 reduces memory and computaional costs by sub-sampling across time, space and…”
- SimVP: Simpler yet Better Video Prediction (2022, methodology): “These methods achieve impressive performance gain by introducing novel neural operators like various RNNs [65–69,71] or transformers [48,70], delicate architectures like autoregressive [21,28,49,55,71] or normalizing flow [73], and applying distinct training strategies such as adversarial training [”