no code implementations • 9 Dec 2023 • Vivek Miglani, Aobo Yang, Aram H. Markosyan, Diego Garcia-Olano, Narine Kokhlikyan
Captum is a comprehensive library for model explainability in PyTorch, offering a range of methods from the interpretability literature to enhance users' understanding of PyTorch models.
no code implementations • NeurIPS 2023 • Fulton Wang, Julius Adebayo, Sarah Tan, Diego Garcia-Olano, Narine Kokhlikyan
We present a method for identifying groups of test examples -- slices -- on which a model under-performs, a task now known as slice discovery.
no code implementations • 28 Mar 2023 • Xuhai Xu, Mengjie Yu, Tanya R. Jonker, Kashyap Todi, Feiyu Lu, Xun Qian, João Marcelo Evangelista Belo, Tianyi Wang, Michelle Li, Aran Mun, Te-Yen Wu, Junxiao Shen, Ting Zhang, Narine Kokhlikyan, Fulton Wang, Paul Sorenson, Sophie Kahyun Kim, Hrvoje Benko
The framework was based on a multi-disciplinary literature review of XAI and HCI research, a large-scale survey probing 500+ end-users' preferences for AR-based explanations, and three workshops with 12 experts collecting their insights about XAI design in AR.
no code implementations • 26 Dec 2022 • Narine Kokhlikyan, Bilal Alsallakh, Fulton Wang, Vivek Miglani, Oliver Aobo Yang, David Adkins
We propose a fairness-aware learning framework that mitigates intersectional subgroup bias associated with protected attributes.
no code implementations • 27 Apr 2022 • David Adkins, Bilal Alsallakh, Adeel Cheema, Narine Kokhlikyan, Emily McReynolds, Pushkar Mishra, Chavez Procope, Jeremy Sawruk, Erin Wang, Polina Zvyagina
We further propose a preliminary approach, called Method Cards, which aims to increase the transparency and reproducibility of ML systems by providing prescriptive documentation of commonly-used ML methods and techniques.
no code implementations • 19 Apr 2022 • Bilal Alsallakh, Pamela Bhattacharya, Vanessa Feng, Narine Kokhlikyan, Orion Reblitz-Richardson, Rahul Rajan, David Yan
We survey a number of data visualization techniques for analyzing Computer Vision (CV) datasets.
no code implementations • NeurIPS Workshop SVRHM 2021 • Bilal Alsallakh, Vivek Miglani, Narine Kokhlikyan, David Adkins, Orion Reblitz-Richardson
When convolutional layers apply no padding, central pixels have more ways to contribute to the convolution than peripheral pixels.
no code implementations • 8 Jun 2021 • Narine Kokhlikyan, Vivek Miglani, Bilal Alsallakh, Miguel Martin, Orion Reblitz-Richardson
Saliency maps have shown to be both useful and misleading for explaining model predictions especially in the context of images.
no code implementations • NAACL 2021 • Hassan Sajjad, Narine Kokhlikyan, Fahim Dalvi, Nadir Durrani
This paper is a write-up for the tutorial on "Fine-grained Interpretation and Causation Analysis in Deep NLP Models" that we are presenting at NAACL 2021.
1 code implementation • 23 Oct 2020 • Vivek Miglani, Narine Kokhlikyan, Bilal Alsallakh, Miguel Martin, Orion Reblitz-Richardson
We explore these effects and find that gradients in saturated regions of this path, where model output changes minimally, contribute disproportionately to the computed attribution.
1 code implementation • ICLR 2021 • Bilal Alsallakh, Narine Kokhlikyan, Vivek Miglani, Jun Yuan, Orion Reblitz-Richardson
We show how feature maps in convolutional networks are susceptible to spatial bias.
2 code implementations • 16 Sep 2020 • Narine Kokhlikyan, Vivek Miglani, Miguel Martin, Edward Wang, Bilal Alsallakh, Jonathan Reynolds, Alexander Melnikov, Natalia Kliushkina, Carlos Araya, Siqi Yan, Orion Reblitz-Richardson
The library contains generic implementations of a number of gradient and perturbation-based attribution algorithms, also known as feature, neuron and layer importance algorithms, as well as a set of evaluation metrics for these algorithms.