Higher-Order Factorization Machines

Authors: Mathieu Blondel, Akinori Fujino, Naonori Ueda, Masakazu Ishihata

NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate the proposed approaches on four different link prediction tasks. 6 Experimental results
Researcher Affiliation Collaboration Mathieu Blondel, Akinori Fujino, Naonori Ueda NTT Communication Science Laboratories Japan Masakazu Ishihata Hokkaido University Japan
Pseudocode Yes Algorithm 1 Evaluating Am(p, x) in O(dm) Algorithm 2 Computing Am(p, x) in O(dm)
Open Source Code No The paper does not provide an explicit statement or a link to the open-source code for the methodology described.
Open Datasets Yes Table 2: Datasets used in our experiments. For a detailed description, c.f. Appendix A. NIPS [17] Enzyme [21] GD [10] Movielens 100K [6]
Dataset Splits Yes We split the n+ positive samples into 50% for training and 50% for testing. We sample the same number of negative samples as positive samples for training and use the rest for testing.
Hardware Specification No The paper mentions running experiments on the NIPS dataset and comparing solvers but does not provide specific details about the hardware (e.g., CPU, GPU models, memory) used for these experiments.
Software Dependencies No The paper mentions `libfm` and `TensorFlow` but does not specify version numbers for any software dependencies used in their experiments or implementation.
Experiment Setup Yes We chose β from 10^-6, 10^-5, . . . , 10^6 by cross-validation and following [9] we empirically set k = 30. Throughout our experiments, we initialized the elements of P randomly by N(0, 0.01). We set ℓ to be the squared loss.