It’s interesting how AI has evolved and brought about some pleasant consequences in the society. The consequences have been so stupendous they continue to amaze us. Such is an example of Nvidia using AI to create pretty believable fake videos.
You might be wondering that similar work has been done before, it’s not something new. Sure, the research is based on an AI method that’s particularly good at generating visual data: a generative adversarial network, or GAN. GANs work by combining two separate neural networks; one that makes the data, and another that judges it; rejecting samples that don’t look accurate.
In this way, the AI teaches itself to generate better and better results over time. This sort of program is common in the industry, and has been used to create all sorts of imagery, from fake celebrity faces to new clothing designs and what not.
“The use of GANs isn’t novel in unsupervised learning, but the Nvidia research produced results — with shadows peeking through thick foliage under partly cloudy skies — far ahead of anything seen before,” the company said.
Nvidia’s research, though, has one big advantage over existing GANS; it learns with much less supervision. Generally, programs of this sort need labelled datasets to generate data. As Nvidia researcher Ming-Yu Liu explained to The Verge, this means that if you’re making a GAN that turns a daytime scene into a nighttime one, you’d need to feed it pairs of images taken at the same location at night and day. It would then study the difference between the two to generate new examples. Pretty cool, isn’t it?
Nvidia has published a blog post to tell of developments in AI and Deep Learning discussed at the Conference and Workshop on Neural Information Processing Systems 2017 – or NIPS for short. The graphics chip maker said that it had two papers accepted to the conference this year, and contributed to two others. Nvidia has over 120 people working on associated AI tech, and plenty of resources, so it’s not surprising it can field so strongly.
The above tech isn’t just useful for transforming scenes quickly and realistically “far ahead of anything seen before.” Nvidia reckons the technology will be useful for self-driving cars, for example. Training data can be captured once and then simulated across a variety of virtual conditions: sunny, cloudy, snowy, rainy, night time, excreta to cover all visual bases for the automobiles in the wild.
This would be a huge leap for the self-driving cars. The information gathered and displayed would help maneuver the cars through various terrains in every weather and seasonal conditions.
You can get detailed information regarding the process and the new technology on the Nvidia Blog, and on the dedicated research page.