no code implementations • 27 Jun 2022 • Helen Lu, Divya Shanmugam, Harini Suresh, John Guttag
Test-time augmentation -- the aggregation of predictions across transformed examples of test inputs -- is an established technique to improve the performance of image classification models.
1 code implementation • 7 Jun 2022 • Angie Boggust, Harini Suresh, Hendrik Strobelt, John V. Guttag, Arvind Satyanarayan
Moreover, with saliency cards, we are able to analyze the research landscape in a more structured fashion to identify opportunities for new methods and evaluation metrics for unmet user needs.
no code implementations • 17 Feb 2021 • Harini Suresh, Kathleen M. Lewis, John V. Guttag, Arvind Satyanarayan
Interpretability methods aim to help users build trust in and understand the capabilities of machine learning models.
no code implementations • 24 Jan 2021 • Harini Suresh, Steven R. Gomez, Kevin K. Nam, Arvind Satyanarayan
To ensure accountability and mitigate harm, it is critical that diverse stakeholders can interrogate black-box automated systems and find information that is understandable, relevant, and useful to them.
no code implementations • 6 Nov 2020 • Alexander D'Amour, Katherine Heller, Dan Moldovan, Ben Adlam, Babak Alipanahi, Alex Beutel, Christina Chen, Jonathan Deaton, Jacob Eisenstein, Matthew D. Hoffman, Farhad Hormozdiari, Neil Houlsby, Shaobo Hou, Ghassen Jerfel, Alan Karthikesalingam, Mario Lucic, Yian Ma, Cory McLean, Diana Mincu, Akinori Mitani, Andrea Montanari, Zachary Nado, Vivek Natarajan, Christopher Nielson, Thomas F. Osborne, Rajiv Raman, Kim Ramasamy, Rory Sayres, Jessica Schrouff, Martin Seneviratne, Shannon Sequeira, Harini Suresh, Victor Veitch, Max Vladymyrov, Xuezhi Wang, Kellie Webster, Steve Yadlowsky, Taedong Yun, Xiaohua Zhai, D. Sculley
Predictors returned by underspecified pipelines are often treated as equivalent based on their training domain performance, but we show here that such predictors can behave very differently in deployment domains.
1 code implementation • 22 May 2020 • Harini Suresh, Natalie Lao, Ilaria Liccardi
ML decision-aid systems are increasingly common on the web, but their successful integration relies on people trusting them appropriately: they should use the system to fill in gaps in their ability, but recognize signals that the system might be incorrect.
no code implementations • 30 Nov 2019 • Ava P. Soleimany, Harini Suresh, Jose Javier Gonzalez Ortiz, Divya Shanmugam, Nil Gural, John Guttag, Sangeeta N. Bhatia
Global eradication of malaria depends on the development of drugs effective against the silent, yet obligate liver stage of the disease.
Cultural Vocal Bursts Intensity Prediction Image Segmentation +1
no code implementations • 28 Jan 2019 • Harini Suresh, John V. Guttag
As machine learning (ML) increasingly affects people and society, awareness of its potential unwanted consequences has also grown.
1 code implementation • 11 Aug 2018 • Willie Boag, Harini Suresh, Leo Anthony Celi, Peter Szolovits, Marzyeh Ghassemi
There are established racial disparities in healthcare, including during end-of-life care, when poor communication and trust can lead to suboptimal outcomes for patients and their families.
1 code implementation • 30 Jun 2018 • Willie Boag, Harini Suresh, Leo Anthony Celi, Peter Szolovits, Marzyeh Ghassemi
In this work, we characterize the doctor-patient relationship using a machine learning-derived trust score.
1 code implementation • 7 Jun 2018 • Harini Suresh, Jen J. Gong, John Guttag
In this work, we present a two-step framework to 1) learn relevant patient subgroups, and 2) predict an outcome for separate patient populations in a multi-task framework, where each population is a separate task.
no code implementations • 23 May 2017 • Harini Suresh, Nathan Hunt, Alistair Johnson, Leo Anthony Celi, Peter Szolovits, Marzyeh Ghassemi
Real-time prediction of clinical interventions remains a challenge within intensive care units (ICUs).
no code implementations • 20 Mar 2017 • Harini Suresh, Peter Szolovits, Marzyeh Ghassemi
We use autoencoders to create low-dimensional embeddings of underlying patient phenotypes that we hypothesize are a governing factor in determining how different patients will react to different interventions.
no code implementations • 16 Dec 2015 • Harini Suresh
Results indicate that the introduced "hill" representation outperforms both the binary and linear representations, the hill representation thus has the potential to improve existing models of ICU mortality.
no code implementations • 12 Jan 2015 • Harini Suresh, Nicholas Locascio
Topic models are a way to discover underlying themes in an otherwise unstructured collection of documents.