Mastering Atari, Go, chess and shogi by planning with a learned model

Nature, Published online: 23 December 2020; doi:10.1038/s41586-020-03051-4

A reinforcement-learning algorithm that combines a tree-based search with a learned model achieves superhuman performance in high-performance planning and visually complex domains, without any knowledge of their underlying dynamics.

Comments

Post a Comment

Popular posts from this blog

Photographing Silhouettes, Part 2

Image
Silhouetted images that pop off the page are graphic in quality, have instantly recognizable subjects and are bold and dominant. Think about the outline of the Grand Teton Range at sunset. The shapes of their peaks are one of the most identifiable ranges in the United States. However, silhouettes don’t have to have bold qualities that show the dominance and stature of a mountain range. A mature weeping willow that stands by a lake’s edge is easily discernible by its form and structure. The feeling it provides is one of softness, tranquility and peace. On the other hand, an outline of a small handgun can instill fear into the viewer. When photographing silhouettes, the size of the subject has nothing to do with how powerful the silhouette can impact the senses. Quite often, a subject photographed as a silhouette can leave a much heavier impression on a viewer than if the same subject was front or side lit. If you are close to my age group, I’m sure you remember the opening scene to t...
Read more