top of page

Machine and Deep Learning 

Machine Learning:

The process in which we manually or intuitively select system or model control features and then train the model. Typical machine learning techniques include decision trees.

Deep Learning:

A subset of machine learning modeled loosely on the neural pathways of the human brain.  In deep learning, the algorithm automatically learns what features are useful. Common deep learning techniques include neural networks.

The relationships between applied Artificial Intelligence Machine Learning and Deep learning can is expressed in the following figure:


Why use Decision Trees over other data mining algorithms?

· Enable one to see the logic unlike black-box methods such as artificial neural networks.

· Easy to implement in a procedural form: IF-THEN-ELSE statements.

· Handle almost any type of data :

· Simplified data preparation. e.g. data normalization not necessary.

The bAware tree building algorithm uses a top-down approach by recursively splitting on a training data set:

· Identifies the feature and splitting criteria that give the best homogeneous sets of data.

· Automatically builds the tree structure by successively creating tree node object instances and associated branches between nodes.

· Splitting stops when all samples in the data set have the same target value a(pure data set).

· A graphical e-presentation of the tree is displayed on a dedicated work-space.

· An executable procedure is generated that contains the tree algorithm.


Graphical Decision Tree Inferred From Historical Data


bAware Decision Tree Add-in Graphical IDE


bAware Artificial Neural Network Add-in Graphical IDE

Why Neural Networks?

Can handle…

  •  high dimensional, multivariate processing units

  • high degree of process non-linearities

  • substantial operational uncertainties

  • large and multiple time delays

In Addition, They....

  • Require minimal process knowledge

  • Are relatively insensitive to process signal noise

  • Are Very Fast


Software Developers can apply neural network technology to predict product parameters in real time, can handle difficult-to-measure quality values through historical data models, and use the intuitive graphical interface  to quickly build and test models

bottom of page