Massive Data Sets Proceedings of a Workshop (1996) / Chapter Skim
Currently Skimming:
Massive Data Sets and Artificial Intelligence Planning
Massive Data Sets and Artificial Intelligence Planning
Pages 105-114
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 105... ...
Examples include personal assistants that filter news stories, schedule meetings, and "crawl the web" in search of information at remote Internet sites. NIany of these applications are easy to build and require little intelligence; the effort of a few hours is enough, for example, to write a filtering program to find calls for papers in the welter of Internet tragic.
|
From page 106... ...
Planning systems can help organize access to enormous quantities of satellite image data A Researchers in information retrieval, machine learning, knowledge discovery in databases, and natural language understanding are finding that the information glut is actually helpful, providing an inexhaustible supply of training data, test cases, and opportunities to develop new techniques.
|
From page 107... ...
By treating the preconditions of these actions as goals to be satisfied in turn, and taking note of potential interactions between actions, the planner recursively generates a sequence of appropriate actions. Traditional Al planners construct plans from primitive actions, starting from scratch for each new problem.
|
From page 108... ...
An action specification is similar to a plan specification, except that its body contains arbitrary code, rather than a control schema. (clef i ne-plan histogram : goal (generate-description : histogram-type ?
|
From page 109... ...
When this happens, the fit is reapplied to the residuals and the line parameters updated appropriately. When the magnitude of the incremental changes falls below some heuristic threshold, the iteration stops.
|
From page 110... ...
In addition to descriptive plans for resistant lines, various box plot procedures, and smoothing procedures, we have implemented forward selection algorithms for cluster analysis and regression analysis [9] , a causal modeling algorithm [8]
|
From page 111... ...
Context constraints apply to the history of plan and subgoal activations that have led up to the current decision point. Using context constraints a rule can, for example, prevent clusters detected in the residuals of a linear fit from being explored if the clustering plan has not yet been applied at the level of the original relationship.
|
From page 112... ...
Our long-term objectives for this research include fully automated model-building and discovery mechanisms driven by an opportunistic control strategy. We expect to develop the automated strategies from our experience with the system as a manual analysis decision aid, letting human analysts provide much of the initial reasoning control strategy.
|
From page 113... ...
The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements either expressed or implied, of the Advanced Research Projects Agency, Rome Laboratory, or the U.S. Government.
|
From page 114... ...
A case study in planning for exploratory data analysis. In Advances in Intelligent Data Analysis, pages 1-5, 1995.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.