Trustworthy Prediction

Trustworthy prediction focuses on developing machine learning models that not only produce accurate predictions but also provide reliable uncertainty estimates and are robust against biases and out-of-distribution data. Current research emphasizes methods for improving calibration, such as novel calibration measures and k-Nearest Neighbor uncertainty estimation, alongside techniques for enhancing fairness and robustness, including causal debiasing and stratified invariance. These advancements are crucial for deploying machine learning models in high-stakes applications like healthcare, finance, and safety-critical systems, where confidence in predictions is paramount. The ultimate goal is to build models that are not only accurate but also transparent and explainable, fostering trust and responsible use of AI.

Papers

July 19, 2024

Truthfulness of Calibration Measures
Nika Haghtalab, Mingda Qiao, Kunhe Yang, Eric Zhao
Truthful Space Sequential Prediction Trustworthy Prediction Calibration Measure

July 2, 2024

Efficient Nearest Neighbor based Uncertainty Estimation for Natural Language Processing Tasks
Wataru Hashimoto, Hidetaka Kamigaito, Taro Watanabe
Uncertainty Estimation Natural Language Inference Nearest Neighbor Natural Language Processing Task Trustworthy Prediction

June 11, 2024

Test-Time Fairness and Robustness in Large Language Models
Leonardo Cotta, Chris J. Maddison
Language Model Native Robustness Causal Reasoning Context Prompting Trustworthy Prediction

May 28, 2024

Towards robust prediction of material properties for nuclear reactor design under scarce data -- a study in creep rupture property
Yu Chen, Edoardo Patelli, Zhen Yang, Adolphus Lye
Native Robustness Robust Learning Robust Prediction Material Property Scarce Data Life Prediction Nuclear Reactor Trustworthy Prediction

May 22, 2024

Counterfactual Gradients-based Quantification of Prediction Trust in Neural Networks
Mohit Prabhushankar, Ghassan AlRegib
Neural Network Supervised ImageNet Gradient Based ImageNet 1k Counterfactual Prediction Image Detection Trustworthy Prediction

January 26, 2024

Inferring Data Preconditions from Deep Learning Models for Trustworthy Prediction in Deployment
Shibbir Ahmed, Hongyang Gao, Hridesh Rajan
Deep Neural Network Deep Learning Model Deep Network DNN Model Software Deployment Model Precondition Trustworthy Prediction

January 23, 2024

Deep Neural Network Benchmarks for Selective Classification
Andrea Pugnana, Lorenzo Perini, Jesse Davis, Salvatore Ruggieri
Machine Learning Model Selective Classification Benchmarking Deep Learning Trustworthy Prediction

June 29, 2023

Integrating Large Pre-trained Models into Multimodal Named Entity Recognition with Evidential Fusion
Weide Liu, Xiaoyang Zhong, Jingwen Hou, Shaohua Li, Haozhe Huang, Yuming Fang
Uncertainty Estimation Large Pre Trained Model Multimodal Named Entity Recognition Trustworthy Prediction Evidential Fusion

January 21, 2023

A Trustworthiness Score to Evaluate DNN Predictions
Abanoub Ghobrial, Darryl Hond, Hamid Asgari, Kerstin Eder
Model Confidence User Trust Suspicious Commenter Trustworthy Prediction Confidence Scoring

April 30, 2022

Explainable Artificial Intelligence for Bayesian Neural Networks: Towards trustworthy predictions of ocean dynamics
Mariana C. A. Clare, Maike Sonnewald, Redouane Lguensat, Julie Deshayes, Venkatramani Balaji
Explainable AI Uncertainty Estimation Bayesian Neural Network Explainable Artificial Intelligence XAI Method Ocean Dynamic Neural Network Parameter Trustworthy Prediction

Trustworthy Prediction

Papers

Truthfulness of Calibration Measures

Efficient Nearest Neighbor based Uncertainty Estimation for Natural Language Processing Tasks

Test-Time Fairness and Robustness in Large Language Models

Towards robust prediction of material properties for nuclear reactor design under scarce data -- a study in creep rupture property

Counterfactual Gradients-based Quantification of Prediction Trust in Neural Networks

Inferring Data Preconditions from Deep Learning Models for Trustworthy Prediction in Deployment

Deep Neural Network Benchmarks for Selective Classification

Integrating Large Pre-trained Models into Multimodal Named Entity Recognition with Evidential Fusion

A Trustworthiness Score to Evaluate DNN Predictions

Explainable Artificial Intelligence for Bayesian Neural Networks: Towards trustworthy predictions of ocean dynamics