Auton Software Applications

See also our Test Programs for simple command-line-interface AI/learning/control software.

FASTMIX

FASTMIX is a very fast implementation of Mixture-model-based EM. It uses the technology described in (Moore, 1999). This is a packaged executable implemented and maintained by Peter Sand. Get FASTMIX here.

NPT

NPT is a very fast implementation of n-point spatial statistics. It uses the technology described in (Gray and Moore, 2000). This is a packaged executable implemented and maintained by Andrew Moore. Get NPT here.

Vizier

Vizier is a complete, integrated Windows95/WindowsNT graphical application for:

• Memory-based Learning (k-nearest-neighbor and kernel regression and classification)
• Locally weighted regression (including locally weighted polynomial regression)
• "Blackbox": Massive cross-validation-based automatic selection of the best learning algorithm for your data.
• Local non-parametric regression analysis
• Graphing/Loading/Saving of fitted learned models
• Automatic generation of C code to implement the model learned from data.
• Non-parametric Experiment Design.

It is based on new kd-tree technology for making the traditionally computationally demanding tasks of locally weighted regression and massive-scale cross validation tractable. It also uses new search algorithms for the blackbox search and memory-based experiment design.

Vizier is a product sold commercially by Schenley Park Research. A demo version is available for evaluation. For details, commercial or non-commercial, contact Jeff Schneider, president of SPR.

Vizier-Excel

Vizier-Excel is a version of Vizier that runs inside the Microsoft(TM) Excel(TM) spreadsheet package. From a research perspective, it provides an interesting example of integrating machine learning in an intuitive interface. For users familiar with Excel it provides an easier way to access the functionality of Vizier (described above).

Vizier-Excel is being developed by Schenley Park Research. For details, commercial or non-commercial, contact Jeff Schneider, president of SPR.

PlanIt

PlanIt is a production scheduling system. It uses stochastic optimization, linear programming, reinforcement learning, cache-based simulation and proprietary search algorithms to optimize stochastic make-to-stock problems for factories which make multiple products simultaneously, in which production statistics vary in a complex, non-linear, fashion and in which costs must be minimized whilst meeting complex by-product demand profiles. For details contact Jeff Schneider, president of SPR.

Return to the Auton Lab Home Page