Epistemic Neural Network

Epistemic neural networks (ENNs) aim to equip machine learning models with the ability to quantify their uncertainty, essentially enabling them to "know what they don't know." Current research focuses on developing ENN architectures, such as those incorporating Bayesian methods or random sets, to improve uncertainty estimation in various applications, including large language models, recommender systems, and anomaly detection. This enhanced uncertainty awareness is crucial for building more robust and reliable AI systems, particularly in high-stakes domains where misclassifications or hallucinations can have significant consequences, leading to improved decision-making and more trustworthy AI outputs.

Papers

April 3, 2024

Composite Bayesian Optimization In Function Spaces Using NEON -- Neural Epistemic Operator Networks
Leonardo Ferreira Guilhoto, Paris Perdikaris
Machine Learning Model Deep Ensemble Operator Learning Function Space Epistemic Neural Network

February 1, 2024

Efficient Exploration for LLMs
Vikranth Dwaracherla, Seyed Mohammad Asghari, Botao Hao, Benjamin Van Roy
Large Language Model Medical LLM Human Feedback Efficient Exploration Epistemic Neural Network

December 25, 2023

Reducing LLM Hallucinations using Epistemic Neural Networks
Shreyas Verma, Kien Tran, Yusuf Ali, Guangyu Min
Large Language Model LLM Hallucination Latent Embeddings Frozen Large Language Model Epistemic Neural Network

July 11, 2023

Random-Set Neural Networks (RS-NN)
Shireen Kudukkil Manchingal, Muhammad Mubashar, Kaizheng Wang, Keivan Shariatmadar, Fabio Cuzzolin
Epistemic Uncertainty R CNN Belief Function Epistemic Neural Network Random Set

June 26, 2023

Scalable Neural Contextual Bandit for Recommender Systems
Zheqing Zhu, Benjamin Van Roy
Recommender System Epistemic Neural Network Neural Contextual Bandit

June 5, 2023

ReContrast: Domain-Specific Anomaly Detection via Contrastive Reconstruction
Jia Guo, Shuai Lu, Lize Jia, Weihang Zhang, Huiqi Li
Contrastive Learning Anomaly Detection Unsupervised Anomaly Detection Medical Anomaly Detection Epistemic Neural Network

February 18, 2023

Approximate Thompson Sampling via Epistemic Neural Networks
Ian Osband, Zheng Wen, Seyed Mohammad Asghari, Vikranth Dwaracherla, Morteza Ibrahimi, Xiuyuan Lu, Benjamin Van Roy
Thompson Sampling Reinforcement Learning Environment Epistemic Neural Network

November 3, 2022

Fine-Tuning Language Models via Epistemic Neural Networks
Ian Osband, Seyed Mohammad Asghari, Benjamin Van Roy, Nat McAleese, John Aslanides, Geoffrey Irving
Language Model Generative Model Language Model Fine Tuning Fine Tuned BERT Epistemic Neural Network

September 5, 2022

Improving Out-of-Distribution Detection via Epistemic Uncertainty Adversarial Training
Derek Everett, Andre T. Nguyen, Luke E. Richards, Edward Raff
Distribution Detection Uncertainty Estimation Distribution Data Epistemic Uncertainty Deep Ensemble Gaussian Noise Epistemic Neural Network

June 15, 2022

Epistemic Deep Learning
Shireen Kudukkil Manchingal, Fabio Cuzzolin
Belief Function Epistemic Neural Network Fuzzy Number

January 27, 2022

Epistemic AI platform accelerates innovation by connecting biomedical knowledge
Da Chen Emily Koo, Heather Bowling, Kenneth Ashworth, David J. Heeger, Stefano Pacifico
Knowledge Graph Digital Innovation Biomedical Research Biomedical Knowledge Epistemic Neural Network Knowledge Map