Lower Dimensional Kernels for Video Discriminators
arXiv:1912.08860 · doi:10.1016/j.neunet.2020.09.016
Abstract
This work presents an analysis of the discriminators used in Generative Adversarial Networks (GANs) for Video. We show that unconstrained video discriminator architectures induce a loss surface with high curvature which make optimisation difficult. We also show that this curvature becomes more extreme as the maximal kernel dimension of video discriminators increases. With these observations in hand, we propose a family of efficient Lower-Dimensional Video Discriminators for GANs (LDVD GANs). The proposed family of discriminators improve the performance of video GAN models they are applied to and demonstrate good performance on complex and diverse datasets such as UCF-101. In particular, we show that they can double the performance of Temporal-GANs and provide for state-of-the-art performance on a single GPU.
References in corpus (13)
- GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
- UCF101: A Dataset of 101 Human Actions Classes From Videos in The Wild
- Large Scale GAN Training for High Fidelity Natural Image Synthesis
- Spectral Normalization for Generative Adversarial Networks
- Self-Attention Generative Adversarial Networks
- Improved Techniques for Training GANs
- Generating Videos with Scene Dynamics
- Which Training Methods for GANs do actually Converge?
- Spatiotemporal Residual Networks for Video Action Recognition
- A Note on the Inception Score
- The Mechanics of n-Player Differentiable Games
- Train Sparsely, Generate Densely: Memory-efficient Unsupervised Training of High-resolution Temporal GAN
- Towards High Resolution Video Generation with Progressive Growing of Sliced Wasserstein GANs
Cited by in corpus (9)
- Generative Adversarial Networks in Computer Vision: A Survey and Taxonomy
- VideoGPT: Video Generation using VQ-VAE and Transformers
- A Good Image Generator Is What You Need for High-Resolution Video Synthesis
- CoReD: Generalizing Fake Media Detection with Continual Representation using Distillation
- CCVS: Context-aware Controllable Video Synthesis
- MotionVideoGAN: A Novel Video Generator Based on the Motion Space Learned from Image Pairs
- FReTAL: Generalizing Deepfake Detection using Knowledge Distillation and Representation Learning
- Latent Neural Differential Equations for Video Generation
- Xp-GAN: Unsupervised Multi-object Controllable Video Generation