High Risk Domain

High-risk domains, such as healthcare and finance, demand AI systems that are not only accurate but also safe, reliable, and interpretable. Current research focuses on improving the performance and trustworthiness of machine learning models in these contexts, exploring techniques like synthetic data generation for training, novel methods for detecting malicious actors (e.g., through domain registration analysis), and the development of inherently interpretable models that leverage domain knowledge and explainable reasoning. This work is crucial for ensuring responsible AI deployment in high-stakes applications, impacting both the development of robust algorithms and the establishment of ethical guidelines for AI development and use.

Papers

October 10, 2024

Evaluating Differentially Private Synthetic Data Generation in High-Stakes Domains
Krithika Ramesh, Nupoor Gandhi, Pulkit Madaan, Lisa Bauer, Charith Peris, Anjalie Field
Language Model Synthetic Data Synthetic Data Generation Private Data Private Synthetic Data Generation High Risk Domain

March 28, 2024

Improving Clinical NLP Performance through Language Model-Generated Synthetic Clinical Data
Shan Chen, Jack Gallifant, Marco Guevara, Yanjun Gao, Majid Afshar, Timothy Miller, Dmitriy Dligach, Danielle S. Bitterman
Generative Model Synthetic Data Human Language Advanced Language Model Synthetic Medical Data Clinical Natural Language Processing Clinical NLP High Risk Domain

January 6, 2024

SecureReg: A Combined Framework for Proactively Exposing Malicious Domain Name Registrations
Furkan Çolhak, Mert İlhan Ecevit, Hasan Dağ, Reiner Creutzburg
Unified Framework Threat Detection Cyber Threat High Risk Domain Malicious Domain

November 25, 2023

Walking a Tightrope -- Evaluating Large Language Models in High-Risk Domains
Chia-Chien Hung, Wiem Ben Rim, Lindsay Frost, Lars Bruckner, Carolin Lawrence
Large Language Model Language Model Natural Language High Risk Domain Agnostic Metric Synthetic Fibre Rope

November 4, 2023

Objectives Are All You Need: Solving Deceptive Problems Without Explicit Diversity Maintenance
Ryan Boldi, Li Ding, Lee Spector
Reinforcement Learning Quality Diversity Algorithm High Risk Domain Diversity Score Deceptive Power

June 4, 2023

(Un)reasonable Allure of Ante-hoc Interpretability for High-stakes Domains: Transparency Is Necessary but Insufficient for Comprehensibility
Kacper Sokol, Julia E. Vogt
Explainable AI Transparency Index Natural Language Description Post Hoc Explainability High Risk Domain Transparent Machine Learning Explanatory Model Human Desire Hoc Interpretability

December 16, 2022

One-shot skill assessment in high-stakes domains with limited data via meta learning
Erim Yanik, Steven Schwaitzberg, Gene Yang, Xavier Intes, Jack Norfleet, Matthew Hackett, Suvranu De
Deep Learning Domain Adaptation Limited Data Competency Assessment Training Domain Skill Representation High Risk Domain Skill Assessment

December 6, 2022

A Learning Based Hypothesis Test for Harmful Covariate Shift
Tom Ginsberg, Zhongyuan Liang, Rahul G. Krishnan
Machine Learning Covariate Shift Distributional Assumption High Risk Domain

October 30, 2022

Distributionally Robust Domain Adaptation
Akram S. Awad, George K. Atia
Domain Adaptation Domain Distribution High Risk Domain Robust Domain Adaptation

August 17, 2022

A Concept and Argumentation based Interpretable Model in High Risk Domains
Haixiao Chi, Dawei Wang, Gaojie Cui, Feng Mao, Beishui Liao
Inherent Interpretability Interpretable Model High Impact Concept Spoken Argumentation Concept Based Explanation High Risk Domain Quantitative Argumentation

July 3, 2022

Interpretable by Design: Learning Predictors by Composing Interpretable Queries
Aditya Chattopadhyay, Stewart Slocum, Benjamin D. Haeffele, Rene Vidal, Donald Geman
Product Design Interpretable Way Post Hoc Explanation Interpretable Knowledge High Risk Domain

High Risk Domain

Papers

Evaluating Differentially Private Synthetic Data Generation in High-Stakes Domains

Improving Clinical NLP Performance through Language Model-Generated Synthetic Clinical Data

SecureReg: A Combined Framework for Proactively Exposing Malicious Domain Name Registrations

Walking a Tightrope -- Evaluating Large Language Models in High-Risk Domains

Objectives Are All You Need: Solving Deceptive Problems Without Explicit Diversity Maintenance

(Un)reasonable Allure of Ante-hoc Interpretability for High-stakes Domains: Transparency Is Necessary but Insufficient for Comprehensibility

One-shot skill assessment in high-stakes domains with limited data via meta learning

A Learning Based Hypothesis Test for Harmful Covariate Shift

Distributionally Robust Domain Adaptation

A Concept and Argumentation based Interpretable Model in High Risk Domains

Interpretable by Design: Learning Predictors by Composing Interpretable Queries