Machine Vision and Applications · 2025
N. Kashani Motlagh, J. Davis, T. Anderson, J. Gwinnup
visionreject optioncalibrationImageNetremote sensing
Invited journal extension of ISVC 2022 Best Paper. Per-class binomial thresholds scale to ImageNet, remote sensing, and long-tailed splits with stronger selective accuracy.
Artifacts: learning-idk
AFRL Technical Report · 2025
N. Kashani Motlagh, T. Anderson, M. Scherreik
LLMsselective predictionabstentionanalyst tools
Instruction-tuned LLMs equipped with abstention heads deliver 8× utility on out-of-distribution tasks.
Artifacts: calibration
WMT 2024 · 2024
N. Kashani Motlagh, J. Davis, T. Anderson, J. Gwinnup, G. Erdmann
multimodal MTvision-language modelsevaluationWMT
Contrastive evaluation shows SOTA multimodal MT models leverage pixels beyond a regularization effect.
Artifacts: calibration
ISVC 2022 · 2022
N. Kashani Motlagh, J. Davis, T. Anderson, J. Gwinnup
visionreject optionselective accuracyImageNet
Binomial modeling of per-class reject thresholds that boost selective accuracy while keeping abstentions calibrated. Extended in MVA 2025 journal version.
Artifacts: learning-idk
ICCV 2021 Workshop on LUAI · 2021
N. Kashani Motlagh, A. Radhakrishnan, J. Davis, R. Ilin
remote sensingdata collectionlabeling pipelineschange detection
Semi-automated scraping plus OpenStreetMap cues to assemble temporal satellite datasets that feed downstream change-detection.
Artifacts: Construction-Site-Satellite-Imagery