
Agent Building and Learning Environment Project
The Agent Building and Learning Environment (ABLE) is a research project
at the IBM T.J. Watson Research Center.
The focus is on building hybrid intelligent agents that include both reasoning
and learning. The ABLE framework consists of a set of core
JavaBeans called AbleBeans and a set of function-specific JavaBeans called
AbleAgents. Application-level agents (JavaBeans) can be constructed
out of these components using the Able Editor or any standard bean builder
environment.
The core AbleBeans include beans for reading and writing data from text
files, for data transformation and scaling using templates, for rule-based
inferencing using boolean and fuzzy logic, and for neural network learning
including back propagation and self-organizing maps. The function-specific
AbleBeans include agents for classification, clustering, prediction, and
genetic search.
Related Links
As part of the ABLE project, we plan to build a FIPA-compliant agent platform
in Java. FIPA, the Foundation for Intelligent
Physical Agents, is an international standards body working toward
agent interoperability.
Version 1.0 of ABLE will soon be available from the IBM alphaWorks
site.
You can download code and documentation from alphaWorks.
Papers
An overview of the ABLE project. This white paper gives
details on the ABLE research project.
Project Members
Joe
Bigus is the project leader. For more information on the ABLE
project, contact bigus@us.ibm.com.
Don Schlosnagle is an ABLE project member and developer.
daschlos@us.ibm.com