Inductive Logic Programming by Unknown

Author:Unknown
Language: eng
Format: epub
ISBN: 9783030492106
Publisher: Springer International Publishing

AUC-ROC

0.923 ± 0.027

0.995 ± 0.004

0.503 ± 0.003

0.500 ± 0.000

0.741 ± 0.016

AUC-PR

0.826 ± 0.056

0.985 ± 0.013

0.356 ± 0.006

0.335 ± 0.000

0.527 ± 0.036

AUC-ROC

0.700 ± 0.186

0.997 ± 0.007

0.968 ± 0.022

0.532 ± 0.019

0.657 ± 0.014

AUC-PR

0.910 ± 0.072

0.992 ± 0.017

0.943 ± 0.032

0.412 ± 0.032

0.658 ± 0.056

5 Conclusion and Future Work

We considered the problem of learning neural networks from relational data. Our proposed architecture exploits parameter tying: instances of the same rule share the same parameters for the same training example. In addition, we explored relational random walks as relational features for training these neural nets. Further experiments on larger data sets could yield insights into the scalability of this approach. Integration with an approximate-counting method could potentially reduce the training time. Finally, understanding the use of such random-walk-based NN as a function approximator can allow for efficient and interpretable learning in relational domains with minimal feature engineering.

Acknowledgements

SN, GK & NK gratefully acknowledge AFOSR award FA9550-18-1-0462. The authors acknowledge the support of Amazon faculty award. KK acknowledges the support of the RMU project DeCoDeML. Any opinions, findings, and conclusion or recommendations expressed in this material are those of the authors and do not necessarily reflect the view of the AFOSR, Amazon, DeCoDeML or the US government.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.