The new method allows the timing of a restart to adapt according to what the search has already learned.
By Shula Rosen
Researchers at Tel Aviv University have developed a new mathematical method that could make computer searches, artificial intelligence systems and scientific simulations faster and more efficient by teaching them when to “start over” during a search.
The research focuses on a concept known as adaptive resetting, in which a search process can restart from the beginning when doing so improves the chances of finding a solution more quickly.
The study, conducted by doctoral students Tommer D. Keidar and Ofir Blumer under the supervision of Prof. Barak Hirshberg and Prof. Shlomi Reuveni at Tel Aviv University’s School of Chemistry, was published in Nature Communications.
The researchers compare the idea to a bee searching for a flower.
If the bee wanders too long without success, returning to the hive and trying a different route may be faster than continuing in the wrong direction. Previous scientific models assumed that such resets happened randomly.
The new method allows the timing of a restart to adapt according to what the search has already learned.
According to the researchers, the framework can be applied to many different problems, including internet searches, artificial intelligence, chemical reactions, protein-folding research and computer simulations.
A key advance is that scientists can now predict how these adaptive restarts will affect a system without having to run large numbers of expensive computer simulations for every possible scenario.
Instead, the method uses a mathematical technique known as “reweighting” to calculate how long a search is likely to take, how different outcomes are distributed, and how a system behaves over time.
The team also showed that machine-learning systems can be trained to determine the best time to restart automatically.
In their study, this significantly accelerated molecular dynamics simulations, including protein-folding calculations that are important in medicine and biotechnology.
According to the researchers, the new approach could help improve search algorithms, reduce computing time and provide scientists with a more efficient way to study complex physical and biological systems.
Do You Love Israel? Make a Donation - Show Your Support!
Donate to vital charities that help protect Israeli citizens and inspire millions around the world to support Israel too!
Now more than ever, Israel needs your help to fight and win the war -- including on the battlefield of public opinion.
Antisemitism, anti-Israel bias and boycotts are out of control. Israel's enemies are inciting terror and violence against innocent Israelis and Jews around the world. Help us fight back!