AIsolver

AI SOLVER STUDIO – USER INTERFACE GUIDE

1.    Main menu

File

• New project
  Create a new project
• Load project
  Load an existing project file
• Save project
  Save current project
• Exit
  Close the application, will stop any active training sessions

Data

• Load testing data
  Load testing data from file
• Process unclassified data
  Run current solution on unclassified data from file

Options

• Performance
  Set program configuration, including number of CPU cores to use and priority of worker threads

Help

• On-line user manual
  Opens AI Solver User Manual from the web
• About
  Displays information about the program

2.    Training data

This section displays information about the loaded training data.

From data file:
Name of file training data was loaded from

Rows omitted due to HindSight:
Number of rows allocated from the training data set to the HindSight feature

Rows for Over-Fitting Prevention:
Number of rows allocated from the training data set to the Over-Fitting prevention feature

Number of data rows:
Number of data rows imported from source data files

Number of training items:
Number of data rows actually used for training. This number is equal to Number of Data Rows minus rows allocated to HindSight and Over-Fitting Prevention.

3.    Testing data

This section displays information about the loaded testing data.

From data file:
Name of file training data was loaded from

Rows omitted due to HindSight:
Number of rows allocated from the testing data set to the HindSight feature

Number of data rows:
Number of data rows imported from source data file

Number of testing items:
Number of data rows actually used for testing. This number is equal to Number of Data Rows minus rows allocated to HindSight.

4.    Problem information

This section displays information regarding the classification problem.

Number of occurring classes:
Number of classes that occur in the training data

Classes grid:
Each row represents one class in the training data, how many times it occurrs and the relative frequenecy

5.    Training process

This section includes controls to start and stop the training process as well as performance information for the current solutions.

Iteration #:
The number of the last training iteration. This number starts at zero when a project is created and grows on each training iteration.

Internal score:
Numerical value that is used to evaluate the fitness of the current solution. This is an internal value that has no direct meaning, except that higher scores indicate better solutions.

Start/Stop Training button:
Pressing this button will start or stop the training process, according to the current state of the program.

Results for current solution grid:
In this grid each row represents one class occurring in the training data and shows how many times it was identified correctly, incorrectly and how reliable the results for this class are in the training data.

Visual progress indicator:
Graph of the internal score value through time, to show visually show the progress of the training.

6.    Testing results

This section shows results from testing, i.e. how the current solution is performing on testing data. This is useful information to determine whether the solution is useful in real-world scenarios.

Update testing results after each training iteration:
If checked this whole section will update with latest information after each training iteration. Some performance can be gained by keeping it unchecked.

Update testing results button:
This button is only enabled when the previous item is unchecked. When pressed, this section will be updated with the latest information.

Testing results for current solution grid:
Each row in this grid represents a class occurring in the testing data as well as how many times it occurrs, how many times it was correctly and incorrectly classified as well as the solutions current reliability for this class.

Statistics on mistakes made with testing data:
Each row in this grid represents a class that has at least one misclassification and what the classification mistake was (i.e. what this class was recognized as).

Popularity: 8% [?]