CALO (Cognitive Assistant the Learns and Organizes) is an intelligent computational assistant for the electronic desktop.


PTIME (Personal Time Manager) is a SPARK agent that assists users in managing their commitments in an adaptive, mixed-initiative and collaborative fashion.

PLIANT (Preference Learning through Interactive, Advisable, Nonintrusive Training) is the schedule preference learning component of PTIME.  It is an SVM learner that uses pairwise preference feedback from a user's natural interactions with PTIME during scheduling to acquire a user preference model in the form of a schedule evaluation function that is used for schedule generation as well as candidate ranking.

LEAPTIME (Learning for PTIME) is the learning package for PTIME.  It comprises PLIANT and three decision tree learners: MIeL (Meeting Importance Learner), PIeL (Person Importance Learner), and LocaL (Location Learner).


LAPDOG (Learning Assistant Procedures from Demonstration, Observation and Generalization) is a learning by demonstration system that provides CALO with the ability to acquire new procedures for accomplishing tasks by generalizing observed traces of the user executing the procedure to be learned.


POIROT is an integrated learning system capable of acquiring complex hierarchical plans from a single example through the coordinated efforts  of a variety of learners, planners, reasoners, and experimenters.  Together with BBN (lead) and USC/ISI, SRI is building the metareasoning component of POIROT to control the learning process.

Last updated: February 24, 2007 AI Center Home Page     Home     Publications