Large-scale Validation of Deep Learning for Logged Bandit Feedback
Paper
The paper of this reseach can be found in the main folder of the repository.
Data
Our experiments are conducted on the Criteo dataset, which can be obtained via (http://www.cs.cornell.edu/~adith/Criteo/).
NeuralBLBF
The contributions of our paper can be found in the NeuralBLBF folder. To train and test a model, the main.py file can be used. The following models are available for usage:
- TinyEmbedFFNN
- SmallEmbedFFNN
- LargeEmbedFFNN
- SparseFFNN
- SparseLinear
- CrossNetwork
POEM and Scripts
The baseline results of the paper: Large-scale Validation of Counterfactual Learning Methods A Test-Bed by Lefortier et al. can be replicated with the POEM and Scripts folder. The code was obtained via the papers related website (http://www.cs.cornell.edu/~adith/Criteo/). The website contains a detailed explanation how the baseline results from the paper can be reproduced.
Authors
- Jasper Adegeest - jasper.adegeest@student.uva.nl
- Verna Dankers - verna.dankers@student.uva.nl
- Michiel van der Meer - michiel.vandermeer@student.uva.nl
- Renzo van Slooten - renzo.vanslooten@student.uva.nl