November 30, 2021

Uk News today – Up to date News, NHS, Health, Sport, Science

For the very latest UK news, with sport, health, science, covid

Nvidia uses AI to recreate Pac-Man on its 40th anniversary

Nvidia uses AI to recreate Pac-Man on its 40th anniversary
No game engine, no rasteriser - everything's generated by a neural network. Nvidia reveals today that it has created a neural network that fully simulates the classic Pac-Man on the event of the Namco coin-op's 40th anniversary. On the face of it, this might not sound like a big deal - Pac-Man is a relatively straightforward…

No video game engine, no rasteriser – everything’s generated by a neural network.

Nvidia reveals today that it has produced a neural network that completely imitates the timeless Pac-Man on the occasion of the Namco coin-op’s 40 th anniversary. On the face of it, this may not seem like a huge offer – Pac-Man is a reasonably straightforward video game that happens in a simplistic, static environment, so using an AI to study its guidelines and completely duplicate its video game logic doesn’t sound outlandishly complex. Except that isn’t what’s happening. There is no engine here, no game reasoning and no traditional rasteriser utilized in the AI entertainment of the game. Rather, everything generated at a per-pixel level is coming straight from the neural network, based upon what it ‘understands’ about how Pac-Man works. This performance of Pac-Man essentially plays out as an AI ‘believes’ it should – and remarkably, it works.

Nvidia is working on something that it describes as GameGAN (GAN meaning ‘generative adversarial network’). It works by utilizing two neural networks working versus one another – a generator and a discriminator. It’s the exact same sort of AI that has actually been utilized thoroughly for a variety of applications, consisting of the creation of AI-generated high resolution texture packs for retro video games.

In the case of Nvidia’s GameGAN, the AI studied 50,000 games of Pac-Man before ‘learning’ how the video game works usually, and how user input impacts what happens on-screen. “This is the very first research to emulate a game engine utilizing GAN-based neural networks,” says.
Seung-Wook Kim, an NVIDIA researcher and lead author on the task. “We desired to see whether the AI might find out the guidelines of an environment just by looking at the movie script of a representative moving through the video game.

Rather than having a human player indulge in 50,000 rounds of Pac-Man, Nvidia trained up a second AI to play through the game instead, offering GameGAN the information it needed to create its own AI rendition.

Running this neural network sets the video game in movement, with each frame produced by the AI’s understanding of the video game – down to each specific pixel in every frame. Some little rendering errors can creep in based on erroneous inferencing, the AI is recreating a Pac-Man that apparently runs just like the original game.

Find Out More