AI News, Natural Video Synthesis with Generative Adversarial Networks ... artificial intelligence

GANs Can Soon Create High Resolution Videos

The primary objective of unsupervised learning is to train an algorithm to generate its own instances of data.

DVD-GAN employs two discriminators for its assessment:  D_S critiques single frame content and structure by randomly sampling k full-resolution frames and processing them individually.

Due to the complexities involved with increased data in case of videos, the generation has been restricted to simple datasets or where strong temporal conditioning information is available.

DVD-GAN which is built upon the state-of-the-art BigGAN architecture introduces a number of video-specific modifications including efficient separable attention and a spatio-temporal decomposition of the discriminator.  Bi-directional Generative Adversarial Networks (BiGANs) were introduced a couple of years ago to learn inverse mapping, and demonstrate that the resulting learned feature representation is useful for auxiliary supervised discrimination tasks, which are on par with unsupervised and self-supervised feature learning.

The above figure Selected frames from videos generated by a DVD-GAN trained on Kinetics-600 at 256 × 256,128 × 128, and 64 × 64 resolutions (top to bottom).

Generating longer and larger videos is a more challenging modeling problem and DVD-GAN is able to generate plausible videos at all resolutions and with actions spanning up to 4 seconds (48 frames).

DeepMind DVD-GAN: Impressive Step Toward Realistic Video Synthesis

The rapid development of AI models such as variational autoencoders (VAE) and generative adversarial networks (GAN) that can generate audio, images and video has opened a Pandora’s box of digital fakery.

In a new paper, UK research company DeepMind introduces DVD-GAN (not “digital versatile disc” but “dual video discriminator”) for video generation on large-scale datasets.

Below is a set of four-second synthesized video clips trained on 12 128×128 frames from Kinetics-600, a large dataset of 10-second, high-resolution YouTube clips originally created for the task of human action recognition.

Although large-scale high-quality data is the fuel that drives machine learning model performance, researchers had struggled to train previous video-generation models efficiently on large datasets due to high data complexity and computational requirements.

DeepMind has overcome this challenge by extending its home-grown image generation model BigGAN to video and introducing extra techniques to accelerate training, including a dual-discriminator architecture consisting of a spatial discriminator and a temporal discriminator, and separable self-attention applied in a row attending over the height, width and time axis.

BigGANs: AI-Based High-Fidelity Image Synthesis

This episode was supported by "Anything outside life extension is a complete waste of time". See their papers: - Papers: ...

Ian Goodfellow: Generative Adversarial Networks (GANs) | MIT Artificial Intelligence (AI) Podcast

Ian Goodfellow is an author of the popular textbook on deep learning (simply titled "Deep Learning"). He invented Generative Adversarial Networks (GANs) and ...

Tutorial on Generative adversarial networks - Visual Synthesis and Manipulation with GANs

ICCV17 | Tutorials | Generative adversarial networks Jun-Yan Zhu, UC Berkeley

End-toEnd Facial Synthesis with Temporal Generative Adversarial Networks

Video showing the generation of talking heads using Generative Adversarial Networks. The method takes as input a still image and an audio clip and produces ...

GauGAN: Changing Sketches into Photorealistic Masterpieces

A deep learning model developed by NVIDIA Research turns rough doodles into highly realistic scenes using generative adversarial networks (GANs). Dubbed ...

StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks

ICCV17 | 1208 | StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks Han Zhang (Rutgers), Tao Xu (Lehigh), ...

Generative Adversarial Networks

Hosted: Rakuten Boston Date: 2018-02-01 Link: Abstract: ..

High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANs


Generative Adversarial Networks (GANs)

A toy example of a simple 2D GAN. See blog post about it here: ...

Deep Learning Papers Review at Neurotechnology: Generative Adversarial Networks

This week we will be presenting Generative Adversarial Networks (GAN) and related developments. You should pay attention to GANs, because it was reported ...