Bayesian Model Selection for Change Point Detection and Clustering

Authors: Othmane Mazhar, Cristian Rojas, Carlo Fischione, Mohammad Reza Hesamzadeh

ICML 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 7. Experimental results. Consider first an experiment based data generated randomly according to the setup of (1) with the same change points of Example 1. This is considered to be an easy case since d m = 4 < d m = 12 N = 2000, which is within the range of signals for which the consistency result of Corollary 6.1 holds. The experiments in Figure 2 show that the algorithm is quite robust to the level of noise as measured by the signalto-noise ratio S/N = magnitude of smallest jump in f σ2 .
Researcher Affiliation Academia 1KTH Royal Institute of Technology, Stockholm, Sweden. Correspondence to: Othmane Mazhar <othmane@kth.se>.
Pseudocode Yes Algorithm 1 Two-Pass Dynamic Programming Algorithm. input data points (yi)N i=1, maximum number of changes D and penalty strength K. 1: y[k,l] := Pl i=k Yi k l + 1 i=k (yi y[k,l])2, 1 k l N.
Open Source Code No The paper does not provide any specific statements about open-source code availability, nor does it include a link to a code repository.
Open Datasets No The paper uses 'simulated data' and mentions 'data generated randomly according to the setup of (1)' but does not provide concrete access information (link, DOI, citation) to a publicly available dataset. It creates its own data for experimentation without making it accessible.
Dataset Splits No The paper does not provide specific details on train/validation/test dataset splits, only stating that 'simulated data' was used.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details, such as library or solver names with version numbers.
Experiment Setup No The paper describes the inputs to Algorithm 1 (maximum number of changes D and penalty strength K) and discusses signal-to-noise ratio in experimental results but does not provide specific experimental setup details such as concrete hyperparameter values, training configurations, or system-level settings used for the simulations.