| Prev | Next | Start of Chapter | End of Chapter | Contents | Index | (2 out of 18)

Introduction

NeurOn-Line provides you with a number of blocks that let you create, manipulate, and operate on vectors.

You can find the Vector Blocks palette under the Data Processing submenu of the Palettes menu:


Choosing When to Evaluate

The attribute Triggering Method lets you choose when a block executes. A block can execute when it receives one new input, all new inputs, or an input from a particular port.

If Triggering Method is... The block evaluates when...
trigger on any
Any of its input ports receives a new value.
trigger on all
All of its input ports receive new values.
trigger on 1
The top input port receives a new value.
trigger on 2
The bottom input port receives a new value.

The following blocks contain this attribute:

Note that in the Vectorizer block, you can only choose trigger on any or trigger on all.

When you reset a block that supports the attribute Trigger Method, NOL treats all the valid values at its input ports as newly received values. If its input values are valid, it will pass a vector the next time you evaluate it, regardless of the setting of Triggering Method.

Creating Vectors and Scalars

These blocks let you create a single vector path out of one or more scalar data paths, or create multiple scalar paths out of one vector data path:

Using Vectors with Classifiers

These blocks are useful when you are using a neural network to classify data:

Manipulating Vectors

These blocks let you combine, split, and reorder vectors:

Inhibiting Vectors

The Vector Inhibit block cuts off the flow of data in a vector path until an inference input port receives a particular value.

Operating on Vector Elements

These blocks let you perform arithmetic operations and other functions on the elements of a vector:

| Prev | Next | Start of Chapter | End of Chapter | Contents | Index | (2 out of 18)

Copyright © 1996, Gensym Corporation, Inc.