Open-world machine learning: A review and new outlooks

F Zhu, S Ma, Z Cheng, XY Zhang, Z Zhang… - arxiv preprint arxiv …, 2024‏ - arxiv.org
Machine learning has achieved remarkable success in many applications. However,
existing studies are largely based on the closed-world assumption, which assumes that the …

SURE: SUrvey REcipes for building reliable and robust deep networks

Y Li, Y Chen, X Yu, D Chen… - Proceedings of the IEEE …, 2024‏ - openaccess.thecvf.com
In this paper we revisit techniques for uncertainty estimation within deep neural networks
and consolidate a suite of techniques to enhance their reliability. Our investigation reveals …

Gradient-regularized out-of-distribution detection

S Sharifi, T Entesari, B Safaei, VM Patel… - European Conference on …, 2024‏ - Springer
One of the challenges for neural networks in real-life applications is the overconfident errors
these models make when the data is not from the original training distribution. Addressing …

Out-of-distribution detection in long-tailed recognition with calibrated outlier class learning

W Miao, G Pang, X Bai, T Li, J Zheng - Proceedings of the AAAI …, 2024‏ - ojs.aaai.org
Existing out-of-distribution (OOD) methods have shown great success on balanced datasets
but become ineffective in long-tailed recognition (LTR) scenarios where 1) OOD samples …

Towards trustworthy dataset distillation

S Ma, F Zhu, Z Cheng, XY Zhang - Pattern Recognition, 2025‏ - Elsevier
Efficiency and trustworthiness are two eternal pursuits when applying deep learning in
practical scenarios. Considering efficiency, dataset distillation (DD) endeavors to reduce …

Online safety analysis for llms: a benchmark, an assessment, and a path forward

X **e, J Song, Z Zhou, Y Huang, D Song… - arxiv preprint arxiv …, 2024‏ - arxiv.org
While Large Language Models (LLMs) have seen widespread applications across
numerous fields, their limited interpretability poses concerns regarding their safe operations …

Revisiting confidence estimation: Towards reliable failure prediction

F Zhu, XY Zhang, Z Cheng… - IEEE Transactions on …, 2023‏ - ieeexplore.ieee.org
Reliable confidence estimation is a challenging yet fundamental requirement in many risk-
sensitive applications. However, modern deep neural networks are often overconfident for …

Test optimization in dnn testing: A survey

Q Hu, Y Guo, X **e, M Cordy, L Ma… - ACM Transactions on …, 2024‏ - dl.acm.org
This article presents a comprehensive survey on test optimization in deep neural network
(DNN) testing. Here, test optimization refers to testing with low data labeling effort. We …

Rcl: Reliable continual learning for unified failure detection

F Zhu, Z Cheng, XY Zhang, CL Liu… - Proceedings of the …, 2024‏ - openaccess.thecvf.com
Deep neural networks are known to be overconfident for what they don't know in the wild
which is undesirable for decision-making in high-stakes applications. Despite quantities of …

Unified classification and rejection: A one-versus-all framework

Z Cheng, XY Zhang, CL Liu - Machine Intelligence Research, 2024‏ - Springer
Classifying patterns of known classes and rejecting ambiguous and novel (also called as out-
of-distribution (OOD)) inputs are involved in open world pattern recognition. Deep neural …