News

The new book
Reactive Search
and Intelligent Optimization
is published by Springer
(November 2008).
Sample pages are visible here.

Reactive Search S.r.l. supports the organization of
LION 3 (Learning and Intelligent OptimizatioN)
Jan 14-18, 2009, Trento, Italy

Reactive Search S.r.l.

Contact us to buy or request information

Reactive Search S.r.l. is a company dedicated to designing and realizing new services based on learning and optimization. In addition, the company assists in the organization of conferences focused on learning and intelligent optimization.

Learning on the Job

Reactive Search is a robust and efficient method for solving difficult optimization problems. The word reactive hints at a ready response to events while alternative solutions are tested. Its strength lies in the introduction of high-level skills often associated to the human brain, such as learning from the past experience, learning on the job, rapid analysis of alternatives, ability to cope with incomplete information, quick adaptation to new situations and events.

The main features of the Reactive Search techniques are:

Learning on the job
Real-world problems have a rich structure. While many alternative solutions are tested in the exploration of a search space, patterns and regularities appear. The human brain quickly learns and drives future decisions based on previous observations. This is the main inspiration source for inserting online machine learning techniques into the optimization engine of Reactive Search.

Rapid generation and analysis of many alternatives
Often, to solve a problem one searches among a large number of alternatives, each requiring the analysis of what-if scenarios. The search speed is improved if alternatives are generated in a strategic manner, so that different solutions are chained along a trajectory in the search space exploring wide areas and rapidly exploiting the most promising solutions.

Flexible decision support
Crucial decisions depend on factors and priorities which are not always easy to describe before starting the solution process. Feedback from the user in the preliminary exploration phase can be incorporated so that a better tuning of the final solutions takes the end user preferences into account.

Diversity of solutions
The final decision is up to you, not the machine. The reason is that not all qualitative factors of a problem can be encoded into a computer program. Having a set of diverse near-best alternatives is often crucial for the decision maker.

Anytime solutions
You decide when to stop searching. A first complete solution is generated rapidly, better and better ones are produced in the following search phases. The more you run, the bigger the possibility to identify excellent solutions, but if you want a solution fast you are going to get it!