Sparse Probabilistic Circuits via Pruning and Growing

Authors: Meihua Dang, Anji Liu, Guy Van den Broeck

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Empirically, our learner achieves state-of-the-art likelihoods on MNIST-family image datasets and on Penn Tree Bank language data compared to other PC learners and less tractable deep generative models such as flow-based models and variational autoencoders (VAEs).
Researcher Affiliation Academia Meihua Dang CS Department UCLA mhdang@cs.ucla.edu Anji Liu CS Department UCLA liuanji@cs.ucla.edu Guy Van den Broeck CS Department UCLA guyvdb@cs.ucla.edu
Pseudocode Yes Algorithm 1: PC sampling
Open Source Code Yes Code and experiments are available at https://github.com/UCLA-Star AI/Sparse PC.
Open Datasets Yes We now evaluate our proposed method pruning and growing on two different sets of density estimation benchmarks: (1) the MNIST-family image generation datasets including MNIST [22], EMNIST [5], and Fashion MNIST [46]; (2) the character-level Penn Tree Bank language modeling task [27].
Dataset Splits Yes We split out 5% of training data as a validation set. We perform early stopping and hyperparameter search using a validation set and report results on the test set.
Hardware Specification No The paper mentions "customized GPU kernels" but does not provide specific hardware details such as GPU models, CPU types, or memory specifications used for the experiments.
Software Dependencies No The paper mentions "Py Torch" and "Juice.jl" but does not specify their version numbers.
Experiment Setup Yes We use hidden Chow-Liu Trees (HCLTs) [25] with the number of latent states in {16, 32, 64, 128} as initial PC structures. We train the parameters of PCs with stochastic mini-batch EM (cf. Section 5). We perform early stopping and hyperparameter search using a validation set and report results on the test set. Please refer to Appendix C for more details.