On Measure Concentration of Random Maximum A-Posteriori Perturbations

Authors: Francesco Orabona, Tamir Hazan, Anand Sarwate, Tommi Jaakkola

ICML 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 4. Experiments We evaluated our approach on a 100 x 100 spin glass model with n = 10^4 variables, for which...
Researcher Affiliation Academia Francesco Orabona ORABONA@TTIC.EDU Toyota Technological Institute at Chicago, 6045 S. Kenwood Ave, Chicago, IL 60637; Tamir Hazan TAMIR@CS.HAIFA.AC.IL Dept. of Computer Science, University of Haifa, 31905 Haifa, Israel; Anand D. Sarwate ASARWATE@ECE.RUTGERS.EDU Rutgers University, Dept. of Electrical and Computer Engineering, 94 Brett Road, Piscataway, NJ 08854; Tommi S. Jaakkola TOMMI@CSAIL.MIT.EDU MIT CSAIL, Stata Center, Bldg 32-G470, 77 Mass Ave. Cambridge, MA 02139
Pseudocode Yes Algorithm 1 Sampling with low-dimensional random MAP perturbations from the Gibbs distribution (Hazan et al., 2013b)
Open Source Code No The paper does not provide any information or links regarding open-source code for the described methodology.
Open Datasets No The paper describes generating data for the experiments ("The local field parameters θi were drawn uniformly at random from [−1, 1]... The parameters θi,j were drawn uniformly from [0, c]") but does not use a publicly available dataset with concrete access information or provide details for its own generated dataset.
Dataset Splits No The paper describes the experimental setup and the generation of data, but does not specify training, validation, or test dataset splits.
Hardware Specification No The paper does not provide any specific details about the hardware used to run the experiments (e.g., GPU/CPU models, memory, or cloud instances).
Software Dependencies No The paper does not provide specific version numbers for any software dependencies used in the experiments.
Experiment Setup Yes We evaluated our approach on a 100 x 100 spin glass model with n = 10^4 variables, for which θ(x1, ..., xn) = Σi∈V θi(xi) + Σ(i,j)∈E θi,j(xi, xj). ... The local field parameters θi were drawn uniformly at random from [−1, 1]... The parameters θi,j were drawn uniformly from [0, c], where c ∈ [0, 4]... We evaluated the expected value of F(Γ) with 100 different samples of Γ.