Tensor Decomposition with Missing Indices

Authors: Yuto Yamaguchi, Kohei Hayashi

IJCAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The experiments on both synthetic and real datasets show that the proposed model achieves higher accuracy in the tensor completion task than baselines.
Researcher Affiliation Academia Yuto Yamaguchi , Kohei Hayashi AIST, Japan yuto.ymgc@gmail.com, hayashi.kohei@gmail.com
Pseudocode Yes Algorithm 1 Variational MAP-EM algorithm for the proposed model with CP decomposition (3rd-order), Algorithm 2 Update-q(in), Algorithm 3 Update-U
Open Source Code Yes Reproducibility. Our code to reproduce the experiments is available at https://goo.gl/W4K86I.
Open Datasets Yes We use the Twitter dataset that is publicly available.1 This dataset consists of about 13M geotagged tweets. ... 1http://noisy-text.github.io/2016/geo-shared-task.html
Dataset Splits No For each index in Dtrain, we generate S samples from Gaussian distribution N( |Xijk, 1)... We randomly divide the samples into training set (70%) and test set (30%). The paper explicitly mentions training and testing sets, but no separate validation set or validation split percentage.
Hardware Specification No The paper does not provide any specific details about the hardware (e.g., GPU, CPU models, memory) used to run the experiments.
Software Dependencies No The paper does not specify any software dependencies with version numbers (e.g., Python, specific libraries, frameworks).
Experiment Setup Yes The size of the tensors is 10 10 10. ... R = 3. ... S = 1 for small dataset, whereas S = 10 for large dataset. ... We vary α from 0 to 0.9 by the step size 0.1. ... Parameters are set to λ = 1.0 and R = 10.