Metadata Dependent Mondrian Processes

Authors: Yi Wang, Bin Li, Yang Wang, Fang Chen

ICML 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We empirically test the proposed MDMP relational model on three real-world data sets with various meta information. We compare MDMP to IRM (Kemp et al., 2006) (block model with Bernoulli distribution in each block) for link prediction, Bi LDA (Porteous et al., 2008) (block model with discrete distribution in each block) for rating prediction, and MP (Roy & Teh, 2009) for both.
Researcher Affiliation Academia Machine Learning Research Group, National ICT Australia, Eveleigh, NSW 2015, Australia School of Computer Science & Engineering, University of New South Wales, Kensington, NSW 2033, Australia
Pseudocode Yes The inference framework for MDMP is outlined in Algorithm 1.
Open Source Code No The paper does not provide any explicit statements about releasing source code or links to a code repository.
Open Datasets Yes The first data set adopted for link prediction is the Lazega s lawyer data (Lazega, 2003). and We adopt a preprocessed data set (Ma et al., 2011), which comprises 21593 users. and We adopt the Movie Lens data set (https://movielens.org/) for rating prediction.
Dataset Splits Yes In our experiments, each data set is partitioned into 5 splits, and each time 4 splits are used for training and the rest one is used for testing. and We randomly select 70 users and 70 items from the entire data set and keep the sparsity of the rating matrix being 80% for evaluation.
Hardware Specification No The paper does not provide specific hardware details (e.g., CPU/GPU models, memory specifications) used for running experiments.
Software Dependencies No The paper does not specify any software dependencies with version numbers.
Experiment Setup Yes For MP and MDMP, we perform 500 iterations of RJMCMC sampling.