Self Hyper-parameter Tuning for Stream Recommendation Algorithms

• Type: Article in International Conference Proceedings Book
• Date: 2019

Authors

Veloso, B (Author) Other; João Gama (Author) FEP; Malheiro, B (Author) Other; Vinagre, J (Author) Other

Conference proceedings International

Title: ECML PKDD 2018 Workshops - DMLE 2018 and IoTStream 2018, Dublin, Ireland, September 10-14, 2018, Revised Selected Papers

Pages: 91-102

Indexing

Scopus - 6 Citations

Other information

Authenticus ID: P-00Q-9G9

DOI: 10.1007/978-3-030-14880-5_8

Abstract (EN): E-commerce platforms explore the interaction between users and digital content ¿ user generated streams of events ¿ to build and maintain dynamic user preference models which are used to make mean-ingful recommendations. However, the accuracy of these incremental models is critically affected by the choice of hyper-parameters. So far, the incremental recommendation algorithms used to process data streams rely on human expertise for hyper-parameter tuning. In this work we apply our Self Hyper-Parameter Tuning (SPT) algorithm to incremental recommendation algorithms. SPT adapts the Melder-Mead optimi-sation algorithm to perform hyper-parameter tuning. First, it creates three models with random hyper-parameter values and, then, at dynamic size intervals, assesses and applies the Melder-Mead operators to update their hyper-parameters until the models converge. The main contribu-tion of this work is the adaptation of the SPT method to incremental matrix factorisation recommendation algorithms. The proposed method was evaluated with well-known recommendation data sets. The results show that SPT systematically improves data stream recommendations.

Language: English

Type (Professor's evaluation): Scientific