Efficient Conditionally Invariant Representation Learning

Authors: Roman Pogodin, Namrata Deka, Yazhe Li, Danica J. Sutherland, Victor Veitch, Arthur Gretton

ICLR 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We conduct experiments addressing two settings: (1) synthetic data of moderate dimension...; and (2) high dimensional image data... We compare performance over all experiments with HSCIC (Quinzan et al., 2022) and GCM (Shah & Peters, 2020).
Researcher Affiliation Collaboration Roman Pogodin Gatsby Unit, UCL rmn.pogodin@gmail.com Namrata Deka UBC dnamrata@cs.ubc.ca Yazhe Li Deep Mind & Gatsby Unit, UCL yazhe@google.com Danica J. Sutherland UBC & Amii dsuth@cs.ubc.ca Victor Veitch UChicago & Google Brain victorveitch@google.com Arthur Gretton Gatsby Unit, UCL arthur.gretton@gmail.com
Pseudocode Yes Algorithm 1 Estimation of CIRCE
Open Source Code Yes Code for image data experiments is available at github.com/namratadeka/circe
Open Datasets Yes d-Sprites (Matthey et al. (2017)) which contains images of 2D shapes generated from six independent latent factors; and the Extended Yale-B Face dataset (Georghiades et al. (2001)) of faces of 28 individuals under varying camera poses and illumination.
Dataset Splits Yes We select the hyper-parameters of the optimizer and scheduler via a grid search to minimize the in-domain validation set loss.
Hardware Specification No The paper discusses the training process and various parameters but does not provide specific hardware details such as GPU/CPU models, processor types, or detailed computer specifications used for running the experiments.
Software Dependencies No The paper mentions software components like 'Adam W (Loshchilov & Hutter (2019)) optimizer' and 'ResNet-18 (He et al., 2016) model pre-trained on ImageNet (Deng et al., 2009)' but does not provide specific version numbers for the programming language, libraries, or other ancillary software dependencies used to replicate the experiments.
Experiment Setup Yes We used Adam (Kingma & Ba, 2015) for optimization with batch size 256, and trained the network for 100 epochs. For experiments on univariate datasets, the learning rate was 1e-4 and weight decay was 0.3; for experiments on multivariate datasets, the learning rate was 3e-4 and weight decay was 0.1.