MIT Polaris AI Algorithm Reduces Web Page Load Times By Up To 34 Percent
Their framework for faster load times is called "Polaris," not to be confused with AMD's forthcoming GPU architecture (they're two completely separate things). What MIT has done is developed a method of mapping out the elements that comprise a webpage so that browsers can more efficiently pull and load the necessary objects that translate into all those visual elements you see.
"The problem is that browsers can’t actually see all of these dependencies because of the way that objects are represented by HTML (the standard format for expressing a webpage’s structure). As a result, browsers have to be conservative about the order in which they load objects, which tends to increase the number of cross-network trips and slow down the page load," the researchers explain.
This is where Polaris comes into play. Polaris automatically tracks all of the interactions between objects and creates a dependency graph for the page. Using that graph, a browser can more efficiently fetch and render the content that makes up a webpage.
James Mickens, a Harvard professor who began working on the project as a visiting professor at MIT in 2014, offers up an example of a traveling businessman. When the businessman visits a city, he might discover more cities that he has to visit before going home. If someone gave him a list of cities beforehand, he could map out the fastest possible route instead of zigzagging his way across the land based on his own discovery.
"For a Web browser, loading all of a page’s objects is like visiting all of the cities," Mickens says. "Polaris effectively gives you a list of all the cities before your trip actually begins. It’s what allows the browser to load a webpage more quickly."