Implicit Convolutional Kernels for Steerable CNNs

Authors: Maksim Zhdanov, Nico Hoffmann, Gabriele Cesa

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

Reproducibility Variable Result LLM Response
Research Type Experimental We prove the effectiveness of our method on multiple tasks, including N-body simulations, point cloud classification and molecular property prediction. In this section, we implement Steerable CNNs with implicit kernels and apply them to various tasks.
Researcher Affiliation Collaboration Maksim Zhdanov AMLab, University of Amsterdam m.zhdanov@uva.nl Nico Hoffmann Helmholtz-Zentrum Dresden-Rossendorf Gabriele Cesa Qualcomm AI Research AMLab, University of Amsterdam
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code Yes Code and data to reproduce all experiments are available on Git Hub.
Open Datasets Yes Dataset. We conduct experiments using the N-body system [23]... Dataset. The Model Net-40 [56] dataset... Dataset. The QM9 dataset [55] is a public dataset...
Dataset Splits Yes From the remaining objects, we take 80% for training and 20% for validation.
Hardware Specification No All the experiments were performed using the Hemera compute cluster of Helmholtz-Zentrum Dresden-Rossendorf and the Iv I cluster of the University of Amsterdam.
Software Dependencies No The paper does not provide specific software dependencies with version numbers.
Experiment Setup Yes For each task, we tune hyperparameters of all models on validation data: the number of layers, the number of channels in each layer, the depth and the width of implicit kernels, and the number of training epochs.