LEVAN –The Self Learning Algorithm for Image Search
We were taught in school days that we live in the age of information. This has now become almost obsolete. It would be right to call it ‘The Age of information Explosion’. Internet and search engines have changed the way we gather information altogether. Often we’re frustrated with the search results that fill our computer screens, worth to be labeled digital junk, for very few of the suggested links would answer our queries befittingly.
Over year with improvisations the algorithms the search engines use have evolved better and better. Yet, we as users aren’t satisfied with its suggestions all the time. Researchers and data miners take it as a challenge forever, trying to better the search results. In our everyday experience we often end up with frustration while we desperately look for images. Basically search engines over years have evolved better at textual data compared to visual data. This owes to the lack of cognition between sensing the required visual data from a textual query. Although Google is offering ‘search by images’ option, wherein you might upload a image and get your search engine gather the information related to that image, it’s still has lot of room for improvements.
Scientists at the University of Washington (UW) and the Programmers in Allen Institute of Artificial Intelligence (AIAI) have together come up with a software, which they call revolutionary in the field of image search in the digital universe, and have proudly named it LEVAN. This is an acronym on expansion gives ‘Learn Everything about Anything’. Learn everything doesn’t mean the users, we might usually think though, refers to the software itself. In that way it would be a self evolving species in the digital planet. That’s the interesting part of the innovation.
Based on the queries for images this software will browse through millions and millions of books and images found online and finds the best ones associated with the word or phrase given in the search string. While asked what the software is all about Ali Farhadi, Assistant professor of Computer Science and Engineering, Washington University has briefed, “ It’s all about discovering associations between visual and textual data”, in his statement. In simple this is a kind of self evolving/ self tutoring visual encyclopedia that takes its feeds from the internet.
If the software does what it claims on paper, it might turn out to be a better educational tool and a one stop resource for requirements like presentations, for which one usually spend hours to find appropriate images. On June 11th a video clip about LEVAN was released in youtube .