The strength of clinical evidence is recoverable from language model representations but not from their stated grades
Paper • 2606.29034 • Published
Soroosh Tayebi Arasteh
soroosharasteh
·
AI & ML interests
None yet
Recent Activity
authored a paper 3 days ago
Self-supervision drives representational convergence in medical foundation models more than clinical supervision authored a paper 3 days ago
CANDOR: Chance-Calibrated Discordance in Frozen Foundation EncodersOrganizations
None yet
authored 3
papers 3 days ago
authored 2
papers about 1 month ago
[bot] Conversion to Parquet
#1 opened 3 months ago
by 
parquet-converter
published a
dataset 3 months ago
authored a
paper 3 months ago
updated a
dataset 3 months ago
authored 11
papers 4 months ago
Case-Grounded Evidence Verification: A Framework for Constructing Evidence-Sensitive Supervision
Paper • 2604.09537 • Published
Diffusion Probabilistic Models beat GANs on Medical Images
Paper • 2212.07501 • Published
SteuerLLM: Local specialized large language model for German tax law analysis
Paper • 2602.11081 • Published • 3
Differential privacy representation geometry for medical image analysis
Paper • 2603.01098 • Published
The role of self-supervised pretraining in differentially private medical image analysis
Paper • 2601.19618 • Published
Agentic retrieval-augmented reasoning reshapes collective reliability under model variability in radiology question answering
Paper • 2603.06271 • Published
Multi-step retrieval and reasoning improves radiology question answering with large language models
Paper • 2508.00743 • Published
Resolution scaling governs DINOv3 transfer performance in chest radiograph classification
Paper • 2510.07191 • Published
Differential privacy enables fair and accurate AI-based analysis of speech disorders while protecting patient data
Paper • 2409.19078 • Published
Differential privacy for medical deep learning: methods, tradeoffs, and deployment implications
Paper • 2506.00660 • Published
Boosting multi-demographic federated learning for chest radiograph analysis using general-purpose self-supervised representations
Paper • 2504.08584 • Published