Bayesian Manifold Learning: The Locally Linear Latent Variable Model (LL-LVM)

Authors: Mijung Park, Wittawat Jitkrittum, Ahmad Qamar, Zoltan Szabo, Lars Buesing, Maneesh Sahani

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

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate the approach on several real world problems. ... 4 Experiments 4.1 Mitigating the short-circuit problem ... 4.2 Modelling USPS handwritten digits ... 4.3 Mapping climate data
Researcher Affiliation Collaboration Gatsby Computational Neuroscience Unit University College London {mijung, wittawat, zoltan.szabo}@gatsby.ucl.ac.uk atqamar@gmail.com, lbuesing@google.com, maneesh@gatsby.ucl.ac.uk Current affiliation: Thread Genius Current affiliation: Google Deep Mind
Pseudocode No No structured pseudocode or algorithm blocks were found.
Open Source Code Yes An implementation is available from http://www.gatsby.ucl.ac.uk/resources/lllvm.
Open Datasets Yes We test our method on a subset of 80 samples each of the digits 0, 1, 2, 3, 4 from the USPS digit dataset... Data were obtained by averaging month-by-month annual precipitation records from 2005 2014 at 400 weather stations scattered across the US (see Fig. 6)... The dataset is made available by the National Climatic Data Center at http://www.ncdc.noaa. gov/oa/climate/research/ushcn/. We use version 2.5 monthly data [15].
Dataset Splits Yes Using the extracted features (in 2D), we evaluated a 1-NN classifier for digit identity with 10-fold cross-validation (the same data divided into 10 training and test sets).
Hardware Specification No No specific hardware details (e.g., CPU, GPU models, memory, or cluster specifications) used for experiments were mentioned.
Software Dependencies No No specific software dependencies with version numbers were mentioned.
Experiment Setup Yes The full EM algorithm starts with an initial value of θ. ... The two steps are repeated until the variational lower bound in Eq. (6) saturates. ... using 9 different EM initialisations. ... For the graph-based methods LL-LVM, LTSA, ISOMAP, and LLE, we used 12-NN with Euclidean distance to construct the neighbourhood graph.