Vol. 8, No 4: 103–109.

Computer Science and Informatics

 2023

Scientific article

UDK 519.226
pdf-version

Arina O. Maslova
master’s degree, Petrozavodsk State University
(Petrozavodsk, Russia),
ari.maslova03@yandex.ru

Markov Chain Monte Carlo Methods for Factorization Machine Learning

Scientific adviser:
Oleg V. Lukashenko
Reviewer:
Roman V. Voronov
Paper submitted on: 11/06/2023;
Accepted on: 11/28/2023;
Published online on: 12/01/2023.
 Abstract. This paper presents a new approach for training factorization machines with the logistic activation function based on a special version of the Gibbs sampling incorporating additional variables with the Polya-Gamma distribution. To evaluate the effectiveness of the proposed method, numerical experiments were performed on both synthetic and real data. The results of these experiments indicate that the developed method can be used for various practical machine learning tasks, including recommendation systems.
Keywords: factorization machines, Polya-Gamma distribution, Gibbs sampling, Bayesian approach

For citation: Maslova, A. O. Markov Chain Monte Carlo Methods for Factorization Machine Learning. StudArctic forum. 2023, 8 (4): 103–109.

References

Chopin N., Ridgway J. Leave Pima Indians alone: Binary regression as a benchmark for Bayesian computation. Statist. Sci, 2017, No. 32 (1), pp. 64–87. DOI: 10.1214/16-STS581.

Freudenthaler C., Schmidt-Thieme L., et al. Bayesian factorization machines. In Proceedings of the NIPS Workshop on Sparse Representation and Low-rank Approximation, 2011, pp. 995-1000.

Polson N.G., Scott J.G., et al. Bayesian inference for logistic models using Polya–Gamma latent variables. Journal of the American Statistical Association, 2013, Vol. 108, Issue 504, pp. 1339-1349.

Rendle S. Factorization machines. In Proceedings of the 10th IEEE International Conference on Data Mining. Sidney, IEEE Computer Society, 2010.

Rendle S., Freudenthaler C., et al. Fast context-aware recommendations with factorization machines. In Proceedings of the 34th ACM SIGIR Conference on Reasearch and Development in Information Retrieval, 2011.

Displays: 262; Downloads: 82;