Table 5.

The desirability of using a hierarchical classification mechanism. This evaluation on test set is conducted by comparing the HIWL with a simpler model SLWSLM (a single layer (model) with weighted sampling and label smoothing). The SLWSLM is designed by removing the hierarchical learning mechanism from HIWL.

HIWLSLWSLM
RecallOverall accRecallOverall acc
CRS0.963396.32 per cent0.976395.90 per cent
IBS0.96280.9579
CSS0.70690.6207
EO0.96160.9335
SPI0.98340.9795
HIWLSLWSLM
RecallOverall accRecallOverall acc
CRS0.963396.32 per cent0.976395.90 per cent
IBS0.96280.9579
CSS0.70690.6207
EO0.96160.9335
SPI0.98340.9795
Table 5.

The desirability of using a hierarchical classification mechanism. This evaluation on test set is conducted by comparing the HIWL with a simpler model SLWSLM (a single layer (model) with weighted sampling and label smoothing). The SLWSLM is designed by removing the hierarchical learning mechanism from HIWL.

HIWLSLWSLM
RecallOverall accRecallOverall acc
CRS0.963396.32 per cent0.976395.90 per cent
IBS0.96280.9579
CSS0.70690.6207
EO0.96160.9335
SPI0.98340.9795
HIWLSLWSLM
RecallOverall accRecallOverall acc
CRS0.963396.32 per cent0.976395.90 per cent
IBS0.96280.9579
CSS0.70690.6207
EO0.96160.9335
SPI0.98340.9795
Close
This Feature Is Available To Subscribers Only

Sign In or Create an Account

Close

This PDF is available to Subscribers Only

View Article Abstract & Purchase Options

For full access to this pdf, sign in to an existing account, or purchase an annual subscription.

Close