A Symbolic Approach to Explaining Bayesian Network Classifiers

Authors: Andy Shih, Arthur Choi, Adnan Darwiche

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

Reproducibility Variable Result LLM Response
Research Type Experimental We consider in particular the compilation of Naive and Latent-Tree Bayesian network classifiers into Ordered Decision Diagrams (ODDs), providing a context for evaluating our proposal using case studies and experiments based on classifiers from the literature.
Researcher Affiliation Academia Andy Shih and Arthur Choi and Adnan Darwiche Computer Science Department, University of California, Los Angeles {andyshih,aychoi,darwiche}@cs.ucla.edu
Pseudocode Yes Algorithm 1 compile-naive-bayes(N), Algorithm 2 compile-latent-tree(N), Algorithm 3 find-mc-explanation(f(X), x), Algorithm 4 pi-cover(f, π), Algorithm 5 pi-inst(f, π, x) are all presented as clearly labeled algorithm blocks.
Open Source Code No The paper does not provide any statements or links indicating that source code for the described methodology is publicly available.
Open Datasets Yes We now consider the Congressional Voting Records (votes) from the UCI machine learning repository [Bache and Lichman, 2013].
Dataset Splits No The paper mentions using a dataset and training a classifier, but it does not provide specific details on training, validation, or test splits (e.g., percentages, sample counts, or k-fold cross-validation setup).
Hardware Specification No The paper does not mention any specific hardware used for running the experiments (e.g., GPU models, CPU types, or cloud computing instances with specifications).
Software Dependencies No The paper does not provide specific version numbers for any software dependencies, libraries, or frameworks used in the experiments.
Experiment Setup Yes To completely specify the naive Bayes classifier, we also need the prior probability of admission, which we assume to be Pr(A=+) = 0.30. Moreover, we use a decision threshold of 0.50, admitting an applicant x if Pr(A=+ | x) .50.