DiSC: Differential Spectral Clustering of Features

Authors: Ram Dyuthi Sristi, Gal Mishne, Ariel Jaffe

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

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate Di SC on a variety of datasets, including MNIST, hyperspectral imaging, simulated sc RNA-seq and task f MRI, and demonstrate that Di SC uncovers features that better differentiate between conditions compared to competing methods.
Researcher Affiliation Academia Ram Dyuthi Sristi, Gal Mishne UC San Diego La Jolla, CA, USA {rsristi, gmishne}@ucsd.edu Ariel Jaffe Hebrew University Jerusalem, Israel ariel.jaffe@mail.huji.ac.il
Pseudocode Yes Algorithm 1 summarizes the steps of the Di SC algorithm.
Open Source Code Yes Code to reproduce the results for Sections 4.1 and 4.2 is available in https://github.com/Mishne-Lab/ Di SC
Open Datasets Yes We evaluate Di SC on a variety of datasets, including MNIST, hyperspectral imaging, simulated sc RNA-seq and task f MRI. The MNIST dataset [23] consists of images of hand written digits from 0 to 9... It has 60k training samples and 10k testing samples. ... Hyperspectral imagery change detection dataset [12]... f MRI dataset consists of 515 subjects... from the Human Connectome Project [39].
Dataset Splits No For MNIST, the paper specifies '60k training samples and 10k testing samples', indicating a train/test split. However, it does not explicitly mention a separate validation split for any of the datasets used.
Hardware Specification No The paper does not specify the hardware used for experiments, such as specific CPU or GPU models. The checklist states: 'Did you include the total amount of compute and the type of resources used (e.g., type of GPUs, internal cluster, or cloud provider)? [No] Our algorithm is not computation expensive.'
Software Dependencies No While code is provided, the paper does not explicitly list specific software dependencies with version numbers (e.g., Python version, library versions like PyTorch, NumPy, etc.) required to reproduce the experiments.
Experiment Setup Yes The choice of the all the hyperparameters (bandwidth, d A and d B) are discussed in Appendix A.