The Impact of Basic Matrix Factorization Refinements on Recommendation Accuracy

Lak, Parisa; Çağlayan, Bora; Bener, Ayşe

doi:10.1109/bdc.2014.19

Cited by 6 publications

(3 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Summary of Regularisation Rate and Learning Rate for each dataset.Another hyper-parameter that should be initialised in matrix factorization models is the number of latent features. This number also depends on the number of observations, number of users and number of items for each dataset[110]. It should be noted that we performed our experiments with 16, 32, 64 latent features and the results for all are approximately similar.…”

mentioning

confidence: 79%

“…There are several hyperparameters that should be set at the initialization of this algorithm. These hyper parameters as explained are regularization rate and learning rate as well as the number of iterations and number of features [110] that should be optimized in the validation phase [33].…”

Section: Scoring Function With Matrix Factorizationmentioning

confidence: 99%

“…In this research, we use 80 and 20 split for training and testing respectively, which is a common practice in machine learning experiments [33]. We also ensured that The hyper-parameters of the Matrix Factorization model in BPR (see 4.3) should be set at the initial step [110]. Two of these parameters are Learning Rate (LR)…”

Section: Choice Of Experiments Designmentioning

confidence: 99%

See 2 more Smart Citations

Personalised ranking with single source implicit information for recommendation tasks a similarity based Monte Carlo Bayesian Personalised Ranking

Parisa¹

2023

Preprint

View full text Add to dashboard Cite

Background: A recommender algorithm’s main goal is to learn user preferences from the user-system interactions and provide a list of relevant items to the user. In information retrieval literature this problem is formulated as learning to rank (LtR) problem. Bayesian Personalized Ranking (BPR) [1] is one of the popular LtR approaches based on pair-wise comparison using single source implicit information. Aim: In this work, we aim to design a recommender system algorithm that generates accurate recommendations. The system should only use a single source implicit user preference information. This is possible through a good approximation of the posterior probability in BPR optimization function. Method: We proposed a Similarity based Monte Carlo approximate solution for the posterior probability in BPR. We used four datasets from different recommendation application domains to evaluate the performance of our proposed algorithm. The input data was pre-processed to match with the requirements of the algorithm. Result: The result of the analysis shows a significant improvement in terms of mean average precision (MAP) for our proposed algorithm compared with the BPR and another alternative extension to BPR. Conclusion: We conclude that the proposed approximate solution is successful in providing the most informative samples to approximate BPR posterior probability. This is confirmed by the significant improvement of the accuracy of the provided ranked list of items for the users. i

show abstract

mentioning

confidence: 79%

Section: Scoring Function With Matrix Factorizationmentioning

confidence: 99%