Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Understanding and Improving Adversarial Robustness of Neural Probabilistic Circuits
Authors: Weixin Chen, Han Zhao
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Empirical results on image classification tasks show that RNPC achieves superior adversarial robustness compared to existing concept bottleneck models while maintaining high accuracy on benign inputs. |
| Researcher Affiliation | Academia | Weixin Chen University of Illinois Urbana-Champaign EMAIL Han Zhao University of Illinois Urbana-Champaign EMAIL |
| Pseudocode | No | The paper describes methods using mathematical formulations and descriptive text but does not include any explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | Yes | The code is available at https://github.com/uiuctml/RNPC. |
| Open Datasets | Yes | We create four image classification datasets. 1) MNIST-Add3: This dataset is constructed from the MNIST dataset [18]... 3) Celeb A-Syn: This dataset is synthesized based on the Celeb A dataset [21]... 4) GTSRB-Sub: This dataset is a subset of the GTSRB dataset [23]... |
| Dataset Splits | Yes | In total, we generate 63,130 images and split them into training, validation, and testing sets by a ratio of 8:1:1. ... In total, we generate 50,000 training images, 10,000 validation images, and 9,990 testing images. ... In total, GTSRB-Sub contains 22,079 training images, 2,759 validation images, and 2,761 testing images. |
| Hardware Specification | Yes | all experiments conducted using eight NVIDIA RTX A6000 GPUs. ... All inference was performed on a single NVIDIA RTX A6000 GPU. |
| Software Dependencies | No | The adversarial attacks used in this paper are implemented using the adversarial-attacks-pytorch library4 [56]. The paper mentions this library but does not provide a specific version number. |
| Experiment Setup | Yes | The attribute recognition model is trained using the sum of cross-entropy losses over all attributes. The training process is conducted with a batch size of 256 for 100 epochs, using the SGD optimizer. |