arXiv ·
MBZUAI researchers have introduced MIRA, a novel framework for improving the factual accuracy of multimodal large language models in medical applications. MIRA uses calibrated retrieval to manage factual risk and integrates image embeddings with a medical knowledge base for efficient reasoning. Evaluated on medical VQA and report generation benchmarks, MIRA achieves state-of-the-art results, with code available on GitHub.
arXiv ·
MBZUAI researchers introduce TerraFM, a scalable self-supervised learning model for Earth observation that uses Sentinel-1 and Sentinel-2 imagery. The model unifies radar and optical inputs through modality-specific patch embeddings and adaptive cross-attention fusion. TerraFM achieves strong generalization on classification and segmentation tasks, outperforming prior models on GEO-Bench and Copernicus-Bench.
arXiv ·
Researchers propose a universal anatomical embedding (UAE) framework for medical image analysis to learn appearance, semantic, and cross-modality anatomical embeddings. UAE incorporates semantic embedding learning with prototypical contrastive loss, a fixed-point-based matching strategy, and an iterative approach for cross-modality embedding learning. The framework was evaluated on landmark detection, lesion tracking and CT-MRI registration tasks, outperforming existing state-of-the-art methods.
MBZUAI ·
MBZUAI researchers presented DEFUSE-MS at MICCAI 2025, a novel AI system for analyzing changes in MRI scans of multiple sclerosis (MS) patients. DEFUSE-MS uses a deformation field-guided spatiotemporal graph-based framework to identify new lesions by reasoning about how the brain has changed. The model constructs graphs of small regions within baseline and follow-up MRIs, linking them across time with edges enriched with learned embeddings of the deformation field. Why it matters: DEFUSE-MS reframes the task from simple "spot the difference" to understanding structural changes, potentially improving the speed and accuracy of MS diagnosis and treatment monitoring.
MBZUAI ·
MBZUAI researchers are introducing MedNNS, a system to be presented at MICCAI 2025, that recommends the best AI architecture and initialization for medical imaging tasks. MedNNS addresses the challenge of inefficient trial-and-error in building medical imaging AI by reframing model selection as a retrieval problem. The system employs a Once-For-All ResNet-like model and a learned meta-space of 720k model-dataset pairs, using dataset embeddings to predict optimal model performance. Why it matters: By automating model selection, MedNNS promises to significantly reduce the time and resources required to develop effective AI solutions for healthcare, particularly in medical imaging.
MBZUAI ·
Researchers from Carnegie Mellon University and MBZUAI have developed a new method called ConceptAligner for precise image editing using AI. The system decomposes text embeddings into independent building blocks called atomic concepts, allowing users to make targeted tweaks without generating entirely new images. Their approach ensures that each latent factor maps to a specific user-controllable dial, enabling accurate concept-level modifications. Why it matters: This research addresses a major limitation in AI image generation, enhancing its usefulness in industries where precise control is crucial, such as advertising and medicine, and improving the reliability of AI-driven creative tools.
arXiv ·
This paper explores language-independent alternatives to morphological segmentation for Arabic NLP using data-driven sub-word units, characters as a unit of learning, and word embeddings learned using a character CNN. The study evaluates these methods on machine translation and POS tagging tasks. Results show these methods achieve performance close to or surpassing state-of-the-art approaches. Why it matters: By offering simpler, more adaptable segmentation techniques, this research can help improve Arabic NLP applications across diverse domains and dialects.
MBZUAI ·
A CMU professor and MBZUAI affiliated faculty presented research on how LLMs store and use knowledge learned during pre-training. The study used a synthetic biography dataset to show that LLMs may not effectively use memorized knowledge at inference time, even with zero training loss. Data augmentation during pre-training can force the model to store knowledge in specific token embeddings. Why it matters: The research highlights limitations in LLM knowledge manipulation and extraction, with implications for improving model architectures and training strategies for more effective knowledge utilization in Arabic LLMs.