Schenley Park Research Inc.

101 Oak Park Place

Mt Lebanon, PA 15243

Phone: 412-268-2339

Fax: 412-268-5571

Email: schneide@cs.cmu.edu
The SOLUTIONS that SPR provide:



SPR's core technology is an integrated collection 
of advanced statistical, machine learning, and 
artificial intelligence algorithms. These methods 
automatically discover general rules and trends 
from underlying data. And then make decisions, 
and invent new controllers, to exploit what they 
discover.



·	Process modelling

·	Experiment design/optimization

·	Multi-format data mining

·	Learning Controllers



Algorithms integrated in SPR approaches 
include



·	Neural networks

·	Parametric and non-parametric 
regression

·	Decision tree learning

·	Reinforcement learning

·	Bayes classifiers

·	Explanation-based learning

·	Memory-based Learning

·	TFIDF Text learning

·	Adaptive Optimal Control

·	Learning Markov Decision 
Processes, HMMs.
The OPPORTUNITIES that SPR 
can exploit for you:

In the past 5 years, manufacturers have sharply 
increased information levels.

·	Data from: Factory, Inventory, Processes

·	Available on-line. Collected real-time.

·	Uses LANs, Databases, Workstations

OPPORTUNITIES:

·	I have all this data. Now what?

·	What can I learn from all these 
maintenance records?

·	What extra strategic information can I 
squeeze from my marketing records?

·	 What can I learn from my   product 
development records?

·	I just instrumented my process line. Now 
what?

·	How can I optimize process yield?

·	How do I get help spotting anomalous 
behavior in my process line?



All these problems have something in 
common. There is substantial data 
exhibiting valuable implicit regularities. 
Regularities that can be automatically 
discovered and made explicit using SPR's 
expertise and software.
SPR EXAMPLES



·	Manufacturing process control

Process models are learned from 
historical data. Advanced non-linear 
learning controllers. In SPR projects to 
date the results have been: Improved 
stability and quality of control. And 
bottom-line improvements in 
manufacturing quality and process cost.

·	Aerospace autonomous 
learning  controllers

Non-linear dynamic systems are 
identified with real-time nonlinear 
nonparametric regression. Optimal 
controllers are refined online with 
reinforcement learning,   trajectory 
planners,  locally linear LQR design. 
SPR ongoing applications include 
helicopter control, aircraft speed control.

·	Detection of database errors

Rules capturing general regularities 
spotted anomalous data entries. SPR 
methods are used to alert operators or 
management of non-standard signals in 
the data. Data driven early warning.

·	Temporal outcome analysis

SPR example: Trends learned from 
patient records enabled predicting 
future health status.

·	Other Example Application 
Areas

·	Textile Manufacture

·	Extrusion Process Control

·	Compound Design Historical Record 
Mining

·	Robotics

·	VLSI process control
WHO WE ARE



Schenley Park Research Inc. was founded in 
1995 by three faculty from Carnegie Mellon 
University with a combined 40 years of 
experience applying automated learning 
algorithms to data analysis problems in the 
industrial, defense, and medical sectors.


SPR EDUCATION AND TRAINING



Each of SPR's founders are experienced teachers 
and have taught university classes, industrial 
tutorials, and invited conference talks covering 
machine learning, neural networks, applications 
in robotics, manufacturing, web information, 
and medical data processing.



SPR can provide  a full range of custom-tailored 
courses for your personnel.















SCHENLEY PARK RESEARCH 
CONTACT INFORMATION:



Schenley Park Research, Inc.

101 Oak Park Place

Pittsburgh, PA 15243



For further information contact 
Jeff Schneider

Email: j.schneider@cs.cmu.edu

Phone: 412-268-2339

Fax: 412-268-5571



Tom Mitchell: t.mitchell@cs.cmu.edu

Phone: 412-268-2611

Andrew Moore: awm@cs.cmu.edu

Phone: 412-268-7599



http://www.cs.cmu.edu/~spr
Tom Mitchell. Vice President. 
Professor of Computer Science 
at CMU. Ph.D., Stanford 1979. 
Author of 80+ papers and two 
books. Has applied AI in robot, 
medical, Web, domains.
Andrew Moore. Vice President. 
Professor of CS & robotics at 
CMU. Ph.D., Cambridge, 1991. 
Author of 30+ papers. Applied AI 
in robot, food process, auto, 
power, domains.
Jeff Schneider. President. Ph.D., 
Rochester, 1994. Has applied AI 
to robot manipulators, food pro-
cessing domains. Has worked at 
GM and TI research labs.
Mary Soon Lee. Chief Program-
mer. MA (Math), Cambridge, MS 
(Aerospace) Cranfield, UK. Five 
years working in AI research 
industry.