Confidence Aware

Confidence-aware systems aim to improve the reliability and robustness of machine learning models by explicitly quantifying and utilizing prediction confidence. Current research focuses on developing methods to estimate confidence at various granularities (e.g., image-wide, instance-level, category-level), incorporating confidence into model training and decision-making processes (e.g., through confidence-weighted fusion, selective prediction, or dynamic thresholding), and using confidence to improve model calibration and address issues like overconfidence and distribution shift. This work is significant because it enhances the trustworthiness and safety of AI systems across diverse applications, from medical image analysis and robotics to natural language processing and advertising.

Papers

March 23, 2023

Confidence-Aware and Self-Supervised Image Anomaly Localisation
Johanna P. Müller, Matthew Baugh, Jeremy Tan, Mischa Dombrowski, Bernhard Kainz
Anomaly Detection Self Supervised Self Supervised Approach Outlier Exposure Confidence Aware Self Supervised Anomaly

March 3, 2023

Dynamic Competency Self-Assessment for Autonomous Agents
Nicholas Conlon, Nisar R. Ahmed, Daniel Szafir
Autonomous Agent Multiple Outcome Human Robot Teaming Competency Assessment Confidence Aware Robot Capability Informed Agent

December 20, 2022

Calibrating Deep Neural Networks using Explicit Regularisation and Dynamic Data Pruning
Ramya Hebbalaguppe, Rishabh Patra, Tirtharaj Dash, Gautam Shroff, Lovekesh Vig
Deep Learning Model Calibration Data Pruning Explicit Regularization Confidence Aware

August 13, 2022

Confidence Matters: Inspecting Backdoors in Deep Neural Networks via Distribution Transfer
Tong Wang, Yuan Yao, Feng Xu, Miao Xu, Shengwei An, Ting Wang
Deep Neural Network Backdoor Attack Formality Transfer Backdoor Trigger Confidence Aware Effective Backdoor Attack

July 18, 2022

The Vocal Signature of Social Anxiety: Exploration using Hypothesis-Testing and Machine-Learning Approaches
Or Alon-Ronen, Yosi Shrem, Yossi Keshet, Eva Gilboa-Schechtman
Environment Exploration Confidence Aware Behavioral Measure Speech Pattern Voice Characteristic Social Anxiety

June 2, 2022

Improving the Robustness and Generalization of Deep Neural Network with Confidence Threshold Reduction
Xiangyuan Yang, Jie Lin, Hanlin Zhang, Xinyu Yang, Peng Zhao
Deep Neural Network Native Robustness Strong Generalization Adversarial Example Adversarial Training Loss Function Better Robustness Confidence Aware Divergence Loss

January 2, 2022

Image Denoising with Control over Deep Network Hallucination
Qiyuan Liang, Florian Cassayre, Haley Owsianko, Majed El Helou, Sabine Süsstrunk
Neural Network External Control Image Denoising Neural Network Output Confidence Aware

November 3, 2021

Confidence Composition for Monitors of Verification Assumptions
Ivan Ruchkin, Matthew Cleaveland, Radoslav Ivanov, Pengyuan Lu, Taylor Carpenter, Oleg Sokolsky, Insup Lee
Closed Loop Confidence Aware High Frequency Display Compositional Capability Assume Guarantee