 |
|
The rapid development of high-speed communication networks has produced an explosion in the number of data sources. As such, new intelligent computer-based methods are required to find and interpret the salient characteristics from data sources. Traditional neural networks produce opaque models that are difficult to interpret. Recent work in ISIS has developed a new transparent, non-linear modelling approach that enables constructed models to be visualised, enhancing their validation and interpretation. The technique combines the representational advantage of a sparse decomposition, with the good generalisation ability of a support vector machine. |
