Minimal Achievable Sufficient Statistic Learning

Authors: Milan Cvitkovic, Günther Koliander

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

Reproducibility Variable Result LLM Response
Research Type Experimental In a series of experiments, we show that deep networks trained with MASS Learning achieve competitive performance on supervised learning, regularization, and uncertainty quantification benchmarks.
Researcher Affiliation Academia 1Department of Computing and Mathematical Sciences, California Institute of Technology, Pasadena, California, USA 2Acoustics Research Institute, Austrian Academy of Sciences, Vienna, Austria.
Pseudocode No The paper does not include any pseudocode or algorithm blocks.
Open Source Code Yes Code to reproduce all experiments is available online.2 https://github.com/mwcvitkovic/ MASS-Learning
Open Datasets Yes We performed all experiments on the CIFAR-10 dataset (Krizhevsky, 2009)
Dataset Splits No The paper mentions 'TRAINING SET SIZE' in tables and 'Test-set classification accuracy', but does not explicitly state a validation set size, percentage, or a specific train/validation/test split.
Hardware Specification No The paper does not provide specific details about the hardware used for the experiments.
Software Dependencies No The paper states 'coded all our models in Py Torch (Paszke et al., 2017)' but does not specify a version number for PyTorch or any other software dependency.
Experiment Setup Yes We use two networks in our experiments. Small MLP is a feedforward network with two fully-connected layers of 400 and 200 hidden units, respectively, both with elu nonlinearities (Clevert et al., 2015). Res Net20 is the 20-layer residual net of He et al. (2015). In all our experiments, the variational distribution qφ(x|y) for each possible output class y is a mixture of multivatiate Gaussian distributions for which we learn the mixture weights, means, and covariance matrices. ... we use a subsampling strategy: we estimate the Jf term using only a 1/|Y | fraction of the datapoints in a minibatch. ... We performed all experiments on the CIFAR-10 dataset (Krizhevsky, 2009), and coded all our models in Py Torch (Paszke et al., 2017). Full details on all experiments is in Supplementary Material 7.7.