I read some Science News issues in quick succession recently and found a few articles, despite the disparities in topics, hit a common theme for me. (Warning: articles may require a subscription.)
* Where steel-belted radials go to die. Researchers used computers to analyze satellite photographs to identify tire dumping grounds, finding some new (not previously known ones).
* Old drug, new trick. The relevant part:
Golub's group is developing a database of how each Food and Drug Administration–approved drug affects the activity, or expression, of about 22,000 human genes.
Researchers already know how gene expression changes in many forms of cancer. By comparing the information on a particular cancer with the gene-expression effects of the entire medical armamentarium, they can potentially identify new therapeutic pairings, Armstrong says.
* Counting on technology to count elephants. Instead of sending people out into the field to spend a lot of time to trying to find and track elephants in order to estimate elephant population, researchers have discovered that computers can do a fairly decent job of identify the sound of elephant footfalls (distinguishing them from the sound of other animals walking) and generate a population estimate that way.
* Estimating a temblor's strength on the fly. Researchers learned that an earthquake's final magnitude can be reasonably accurately estimated from the first seconds of a quake, enabling an earlier warning system.
What do all these articles have in common? Scientists have used simple computer algorithms (mostly machine learning and data mining ones) to solve useful previously unsolved or unexplored problems.
Posts
No comments:
Post a Comment