Concrete-NN

Data is private and belongs to ETS and uWaterloo Concrete Teams. Contact if you wish to request access.

UCI Folder

Simple neural network based on the University of California Irving Concrete dataset to predict concrete compressive strength.

ETS Folder

Stacked Auto-Encoder based on the ETS ultra light-weight concrete dataset. Used as a sparse recommendation system.

Network Diagram

Training/Testing Cost Histogram

Cost Function Over Num. Epochs

TODO

Implement more complex models in Keras

Citation

Ouellet, J., Martel, J. L., Ouellet-Plamondon, C., & Carter, A. (2016). Predicting the compressive strength of ultralightweight concrete by an artificial neural network.

Acknowledgements

ETS Concrete Canoe Team for data.