A Review of ALNBench by Dendronic Systems Inc. Bruce Matichuk Shengjiu Wang

Overview 4 What is ALNBench? –Motivation/Applications 4 Theory of Multidimensional Piecewise Linear Regression 4 Components of ALNBench 4 Demonstration 4 Questions

What is ALNBench 4 Automated Learning Nets 4 Tool that automates the creation of multidimensional linear regression models 4 Facilitates use of “learned” model 4 Also contains API for building “learning” applications

MPLR Model 4 Consists of a set of linear pieces MAX (a, b, c, d) defines convex-down function MAX (MIN (a, b, c), MIN (d, e, f, g)) defines a function with two bumps

Neural Network Advantages Over Direct Programming 4 Less time to build solution 4 Easier –less code –less setup 4 Ability to solve hard problems

Advantages of ALNs Over Neural Networks 4 Safety 4 Speed 4 Scalability 4 Broad domain applicability 4 Embedding expert knowledge 4 Function inversion 4 Ease of understanding

Other Advantages Include 4 Uses arbitrary ranges of data values.  Grows the architecture automatically.  Allows the user to easily set the parameters.  Trains and evaluates at high speed.  Tells the user the sensitivity of the output to each input in different parts of the input space.  Gives the user unprecedented control over the learned function

Applications 4 Analysis 4 Prediction 4 Control

Views 4 Workspace View 4 Charts View 4 Report View 4 LF Statistics View 4 Log View

Steps to Using ALNBench 4 Create the ALN 4 Load in the training and test data 4 Set the ALN training parameters 4 Train the ALN 4 Examine the results 4 Test other data sets against learned model 4 Save the.alb file

Demonstration 4 United States Total Stock Returns,