|
|
Comparison between BrainMaker and Professional
There are six basic differences between standard BrainMaker ($195) and BrainMaker Professional ($795):
- Building your networks: Bigger networks
With Standard, you're limited to 512 neurons per layer and 32K connections per layer. "Neurons per layer" is the same thing as "independent variables". With Pro, you can have up to 32K neurons per layer and there's no limit on connections except that the network has to fit in your computer's memory.
- Training your networks: Automated parameter changing
Pro lets you use different network parameters for each layer, and lets you set up ways for parameters to change in response to how your network is training. It also includes R-squared calculations.
- Analyzing your networks: Determining input importance
It's common to want to know how a trained network is coming up with the responses it's giving you. Professional lets you "open up the black box" and play "what-if" games to see which of your inputs makes the most difference to your output and under what conditions.
- Distributing your networks: RUNTIME.C
Pro includes a runtime license and source code to permit you to incorporate a trained network in a program you write and distribute.
- Special Financial Forecasting Features
BrainMaker Professional has these special features designed to make predictions easy and rewarding:
- Seven types of analysis including Data Correlator, Cyclic Analysis, Sensitivity Analysis, Global Network Analysis, Contour Analysis, Neuron Sensitivity and NetChecker.
- Automated historical inputs (recurrence).
- Indicators built-in: RSI, MACD, Stochastics, On-Balance Volume, and 6 moving averages.
- Reads CSI, MetaStock, SmartTrader (CompuTrack) files as well as Excel, Lotus 1-2-3, dBase, ASCII and binary.
- Historical financial databases available with automatic file creation and interpolation for BrainMaker included
- Designing your networks: Competitor
Pro has an additional network design program intended for horseracing and similar sorts of problems. Competitor is so simple, it really is just "type in your data, press a button, and it goes".
Feature List for BrainMaker and Professional
User Interface
| Feature
| Standard
| Pro
| Benefit
|
| Pull-down Menus, Dialog Boxes
| X
| X
| easy to learn and use; all parameters saved in a file you can
edit.
|
| Programmable Output Files
| X
| X
| exports data in your format to spreadsheets, graphics
packages, etc.
|
| Editing in BrainMaker
| X
| X
| quickly edit data, display, network connections, and more.
|
| Network Progress Display
| X
| X
| monitors training with a simple graphic display.
|
| Fact Annotation
| X
| X
| attaches your comments to examples for display and printing.
|
| Dynamic Data Exchange (DDE)
| X
| X
| puts your trained network into other Windows programs.
|
| Graphics Built In
|
| X
| see trends, cycles, network responses, statistics, etc.; make
plots.
|
Performance
| Feature
| Standard
| Pro
| Benefit
|
| Binary Mode
| X
| X
| uses binary files for greater speed.
|
| Batch Mode
|
| X
| add networks to your existing programs; train while you're
away.
|
| EMS and XMS Memory
| X
| X
| up to 512 or 8192 independent variables.
|
| Save Network Periodically
| X
| X
| saves results to a file in case of power failure.
|
| Fastest Algorithms
| X
| X
| 43,000,000 connections-per-second (PII/300)
|
| Neurons per Layer
| 512
| 8192
| model complex data with ease; up to 37,767 with Windows.
|
| Number of Layers
| 8
| 8
| extra hidden layers can help tackle bigger problems.
|
Training
| Feature
| Standard
| Pro
| Benefit
|
| Specify Parameters by Layer
|
| X
| fine-tunes performance inside the network four different
ways.
|
| Recurrent Networks
|
| X
| puts feedback in your network and automates past input data.
|
| Prune Connections and Neurons
|
| X
| improves accuracy by trimming away excess "fat".
|
| Heavy Weights
|
| X
| helps networks requiring many training iterations.
|
| Add Neurons While Training
| X
| X
| finds best size network quickly; fully automated with
Professional.
|
| Custom Neuron Functions
| X
| X
| optimizes training to suit any need.
|
| Testing While Training
| X
| X
| trains for best performance on new data
|
| Stop Training When...
| X
| X
| lets you decide when the network has learned well.
|
| Input noise, blurring, symmetry
| X
| X
| creates networks that generalize better.
|
| Hypersonic Training
|
| X
| trains faster with this proprietary algorithm.
|
Analysis Tools and Advanced Functions
| Feature
| Standard
| Pro
| Benefit
|
| Sensitivity Analysis
|
| X
| shows you which inputs determined your results.
|
| Neuron Sensitivity
|
| X
| shows you the total effect of one input on your results.
|
| Global Network Analysis
|
| X
| reports how your network reacts to all your facts overall.
|
| Contour Analysis
|
| X
| shows you color peaks and valleys of response to pairs of
inputs.
|
| Data Correlator
|
| X
| finds important data and optimum time delays.
|
| Error Statistics Report
| X
| X
| check your network error rate during training.
|
| Print or Edit Weight Matrices
| X
| X
| examine and customize network internals.
|
| Competitor
|
| X
| ranks horses, teams, stocks, etc. in finish order.
|
| Run Time System
|
| X
| C source code - make programs with your network for resale.
|
| Genetic Training Option
|
| G
| trains variations of your design and shows you which was the
best.
|
Network Data Management Functions
| Feature
| Standard
| Pro
| Benefit
|
| NetMaker
| X
| X
| spreadsheet-like data manipulation and network file creation.
|
| NetChecker
| X
| X
| checks your files for errors and inconsistencies.
|
| Shuffle
| X
| X
| mixes up the order of examples for better training.
|
| Binary
| X
| X
| converts files to binary for quicker training.
|
| MinMax
| X
| X
| finds min / max / standard deviation of data for fine-tuned
results.
|
| Data Importation
| X
| X
| reads data from Lotus, dBase, Excel, ASCII, binary.
|
| Financial files
|
| X
| reads MetaStock, CompuTrack and CSI financial files.
|
| Data Manipulation
| X
| X
| performs 28 row/column operations; finds outliers; converts
symbols.
|
| Financial Indicators
|
| X
| creates five types of financial indicators; finds trends.
|
| Cyclic Analysis
|
| X
| checks data for periodic or cyclic behavior.
|
| Data Types
| X
| X
| uses symbolic, text, picture, and numeric data.
|
|
