| EPSRC Reference: |
EP/D002540/1 |
| Title: |
MASSING: Multi-Agent Search Strategies in Natural Groups |
| Principal Investigator: |
Holcombe, Professor WML |
| Other Investigators: |
|
| Researcher Co-Investigators: |
|
| Project Partners: |
|
| Department: |
Computer Science |
| Organisation: |
University of Sheffield |
| Scheme: |
Standard Research (Pre-FEC) |
| Starts: |
01 October 2005 |
Ends: |
30 September 2008 |
Value (£): |
160,037
|
| EPSRC Research Topic Classifications: |
| Fundamentals of Computing |
Information & Knowledge Mgmt |
| New & Emerging Comp. Paradigms |
Population Ecology |
|
| EPSRC Industrial Sector Classifications: |
| No relevance to Underpinning Sectors |
|
|
| Related Grants: |
|
| Panel History: |
|
|
Summary on Grant Application Form |
|
Modern computer systems are extremely powerful but often fail because they cannot cope with the constantly changing data they must deal with. As computers get bigger, and the amount of data they manage increases, performance declines because of the difficulties of dealing with a constant changing system. In fact the ability of computer users to get useful information out of large and ever-changing data sets is hindered by the lack of software to quickly search data. Computer scientists think the answer is to develop computer software with many individual programs or 'agents' working together to solve these problems. Using lots of agents is very effective and we only need look at Nature to see that working together gets things done faster and better. Looking at ants has inspired a scientist called Marco Dorigo to develop computer programs called 'ant algorithms' that solve search problems using pheromones, just like real ants. However, Dorigo admits that his ant algorithms don't behave just like ants. Ants provided him with an idea but the problem he wanted to solve was different, so he made them work to solve computing problems. Many species other than ants work together to find resources, but they often do it differently to the ants (Lasius niger) that inspired Marco Dorigo. Obviously, animal species solve different problems because they eat different things and live in different places. However, we believe that if we can work out what problems these animals are solving then we can learn why they have evolved a particular way of finding food. Then, if we can find a computer science problem that is similar we would understand how to make our agents behave in solving that problem. A colony of honeybees searching for nectar shares many similarities with computer agents searching for information in a network. The world is changing and more business is done over the Internet so it will be valuable to have an army of agents at your disposal to find resources at the best price or find buyers for your product. It is also easy to imagine agents routing phone calls in a telecommunications network.Our project will make computational models of six animal species, which have been well-studied. These species work together to find food and live together in one place. Some species like snails and caterpillars live in places with lots of food, whilst others like honeybees have to search large spaces for nectar. Another species we will model is an ant that lives in the desert, where food is very rare. With the models we make we can ask questions about the problems each species must solve to survive. Does their food last a long time? Do they have to search a large area? Do they use a lot of individuals to find food? How do individuals communicate? The answers to these questions will show us the best thing to do when searching for a solution, whether we need to find food or information. Once we know how to identify these problems we can design individual agents that work together and find the answers we need. In fact, because we know how natural agents behave individually and what happens when they work together we can be sure that what they will do is what we expect. If we just guessed how agents should work together then we cannot be sure what will 'emerge' when the agent's interact. It could be that when we put them together they will do something unexpected. This would be dangerous if they were managing a power network, for example. Using 'natural' agents we can be sure of what will happen.By copying Nature we are sure to find the best way of solving very complicated problems with agents. If we choose to guess about how agents should work together we could make very costly, or even dangerous, errors. Understanding natural solutions arrived at over millions of years of evolution will show us how to build useful communities of agents that can work together in harmony.
|
|
Key Findings |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Potential use in non-academic contexts |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Impacts |
|
| Description |
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk |
| Summary |
|
| Date Materialised |
|
|
|
|
Sectors submitted by the Researcher |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
| Project URL: |
|
| Further Information: |
|
| Organisation Website: |
http://www.shef.ac.uk |