news

Aug 09, 2026 Both VLM preprints are now accepted as ECCV 2026 workshop papers: Attributes Should Come from Images, Not Class Names at the PFATCV Workshop, and Ask Twice, Look Twice at the eXCV Workshop. :tada:
Jul 21, 2026 Two new preprints on VLMs: Attributes Should Come from Images, Not Class Names (distribution-conditioned attribute selection that beats prompt tuning) and Ask Twice, Look Twice (prompt echoing resolves the question-first paradox, up to +19 points training-free). :page_facing_up:
Jun 15, 2026 DetPO: In-Context Learning with Multi-Modal LLMs for Few-Shot Object Detection was accepted at ECCV 2026! State-of-the-art few-shot detection with black-box VLMs, outperforming specialist detectors like GroundingDINO. :fire:
May 15, 2026 Our paper Activation Reward Models for Few-Shot Model Alignment was accepted to ACL 2026 Findings! :sparkles:
May 11, 2026 Started my summer research internship at Bosch Research (Robert Bosch LLC) in Pittsburgh, working on grounding vision-language-action (VLA) models for autonomous-driving applications, advised by Madan Ravi Ganesh and Chaithanya Kumar Mummadi. :car:
Feb 16, 2026 I successfully completed my Ph.D. proposal and am now a Ph.D. candidate. I am on the job market for research scientist roles (graduating late 2026); here is my resume. :rocket:
Jan 20, 2026 Three papers accepted at IEEE ISBI 2026: LEARNER (contrastive learning of granular labels), Learning Generic Lung Ultrasound Biomarkers, and Label Uncertainty for Ultrasound Segmentation (Oral)! :tada:
Sep 01, 2025 Honored to receive the Liang Zhao Endowed Fellowship at CMU. :trophy:
Apr 01, 2025 Presented Weakly Supervised Contrastive Learning for Better Severity Scoring of Lung Ultrasound at IEEE ISBI 2025. :sparkles:
Dec 26, 2024 新智元 (New Zhiyuan) covered our ICML 2024 Oral paper, “CMU and others propose a highly practical ‘model generalization’ metric, improving out-of-distribution accuracy by 6%”, on LCA-on-the-Line. :newspaper:
Jun 01, 2024 Three U.S. patent applications published: probabilistic confidence labels (US20240177000A1), direct diagnostic semantic segmentation (US20240177445A1), and imaging-biomarker feature extraction (US20240249498A1). :page_with_curl:
May 02, 2024 Our paper LCA-on-the-Line: Benchmarking Out-of-Distribution Generalization with Class Taxonomies was accepted at ICML 2024 as an Oral! :sparkles:
Jun 01, 2023 Presented Improving Model’s Interpretability and Reliability using Biomarkers at the Bern Interpretable AI Symposium (BIAS 2023). :speech_balloon:
Jul 01, 2022 My lung ultrasound AI research was featured in CMU School of Computer Science’s LINK Magazine: Seeing Beneath The Surface. :newspaper:
Mar 01, 2022 Awarded the CMLH Translational Fellowship in Digital Health at CMU. :trophy:
Feb 01, 2022 Our paper W-Net: Dense and diagnostic semantic segmentation of subcutaneous and breast tissue in ultrasound was published in the Medical Image Analysis journal! :sparkles:
Aug 01, 2021 Started my Ph.D. in the Robotics Institute at Carnegie Mellon University, advised by Prof. John Galeotti and Prof. Deva Ramanan. :mortar_board:
Jul 16, 2021 Our paper The Role of Pleura and Adipose in Lung Ultrasound AI was accepted at the LL-COVID19 workshop, MICCAI 2021! :sparkles:
Jun 01, 2021 Started my research internship at DawnLight Inc., working on multimodal human action recognition for medical and assisted-living settings, with manager Hadi Kiapour, advised by Prof. Fei-Fei Li and Prof. Ehsan Adeli.