Learning about an exponential amount of conditional distributions

Authors: Mohamed Belghazi, Maxime Oquab, David Lopez-Paz

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Throughout a variety of experiments on synthetic and image data, we show the efficacy of NCs in generation and prediction tasks (Sections 5 and 7).
Researcher Affiliation Collaboration 1Facebook AI Research, Paris, France 2Montréal Institute for Learning Algorithms, Montréal, Canada
Pseudocode No The paper describes the training process in six steps but does not provide structured pseudocode or an algorithm block.
Open Source Code No The paper does not provide an explicit statement about releasing source code or a link to a code repository.
Open Datasets Yes We train NCs on SVHN and Celeb A. We consider data imputation tasks on three UCI datasets [37]. [37] Moshe Lichman et al. Uci machine learning repository, 2013.
Dataset Splits Yes We train 40 linear SVMs on learned representations extracted from the encoder using full available and requested masks (a = r = 1) on the Celeb A validation set.
Hardware Specification No The paper does not provide specific details about the hardware used for experiments, such as GPU models or CPU specifications.
Software Dependencies No The paper mentions the 'Adam optimizer' but does not specify version numbers for any software dependencies.
Experiment Setup Yes We train the networks for 10, 000 updates, with a batch-size of 512, and the Adam optimizer with a learning rate of 10 4, β1 = 0.5, and β2 = 0.999. For these experiments, both the discriminator and the NC have 2 hidden layers of 64 units each, and Re LU non-linearities.