Sparse Variational Inference for Generalized GP Models

Authors: Rishit Sheth, Yuyang Wang, Roni Khardon

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

Reproducibility Variable Result LLM Response
Research Type Experimental An experimental evaluation for count regression, classification, and ordinal regression illustrates the generality and advantages of the new approach.
Researcher Affiliation Collaboration Rishit Shetha RISHIT.SHETH@TUFTS.EDU Yuyang Wangb WANGYUYANG1028@GMAIL.COM Roni Khardona RONI@CS.TUFTS.EDU a Department of Computer Science, Tufts University, Medford, MA 02155, USA b Amazon, 500 9th Ave N, Seattle, WA, USA
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code No The paper mentions using third-party tools like the 'GPML toolbox' and 'min Func software' but does not state that the authors are releasing their own code for the methodology described.
Open Datasets Yes The dataset ucsdpeds1l (Chan & Vasconcelos, 2012) ... The remaining datasets are available from the UCI Machine Learning Repository (Lichman, 2013).
Dataset Splits Yes For a given set size, the subset/active set is randomly selected from the data without replacement. After this set is selected, 10fold cross validation is performed with the remaining data.
Hardware Specification No Some of the experiments in this paper were performed on the Tufts Linux Research Cluster supported by Tufts Technology Services. This statement indicates a computing environment but lacks specific details such as CPU/GPU models, memory, or number of cores.
Software Dependencies No The paper mentions using the 'GPML toolbox' and 'min Func software' but does not specify their version numbers, which is required for a reproducible description of software dependencies.
Experiment Setup Yes A Gaussian RBF kernel is used in all cases. A zero-mean function is used in all cases except count regression where a constant mean function is used. ... The hyperparameters are either estimated from the active set or set to default values (σ2 = 1) prior to training using the same procedure across methods. ... Stopping conditions are f(xk) 10 5, f(xk 1) f(xk) 10 9, or k > 500 where f is the objective function being optimized, k represents the iteration number, and x is the current optimization variable.