Pooling Image Datasets with Multiple Covariate Shift and Imbalance

Authors: Sotirios Panagiotis Chytas, Vishnu Suresh Lokhande, Vikas Singh

ICLR 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We show the effectiveness of this approach via extensive experiments on real datasets. Further, we discuss how this style of formulation offers a unified perspective on at least 5+ distinct problem settings, from self-supervised learning to matching problems in 3D reconstruction. The code is available at https://github.com/SPChytas/CatHarm. and 5 EXPERIMENTAL EVALUATIONS
Researcher Affiliation Academia Sotirios Panagiotis Chytas UW-Madison Vishnu Suresh Lokhande UW-Madison Vikas Singh UW-Madison
Pseudocode Yes Algorithm 1 Structure preserving training of Functors
Open Source Code Yes The code is available at https://github.com/SPChytas/CatHarm.
Open Datasets Yes The ADNI dataset can be obtained from https://adni.loni.usc.edu/. and Besides the medical image datasets, we evaluate our performance in two tabular datasets; the German(Hofmann, 1994) and the Adult Becker & Kohavi (1996).
Dataset Splits Yes All the experiments are a result of 5-fold cross validation procedure.
Hardware Specification No The paper does not provide specific details about the hardware (e.g., GPU models, CPU types, or memory) used for running the experiments.
Software Dependencies No The paper mentions using a 'modified Res Net' and 'Fully-Connected Neural network' but does not specify any software dependencies with version numbers (e.g., Python, PyTorch, TensorFlow versions).
Experiment Setup Yes We model the Functor F using a modified Res Net (He et al., 2016), and the Functor C using a Fully-Connected Neural network. For our experiments, using linear mappings W Rn n for the Morphisms in the target Category S (i.e., latent space) was sufficient but this can be easily upgraded. All the experiments are a result of 5-fold cross validation procedure. and In this experiment, we set n = 128. and In this experiment we set n = 32. and moderate values of λ (i.e. 0.01) lead to a low MMD value