ASR
Not all transcription mistakes are equal. In clinic visits, the go-to score Word Error Rate (WER) can look fine even when errors change care-critical meaning. * Expert clinicians compared true utterances with ASR transcripts in two doctor-patient datasets, labeling the impact: No, Minimal, or Significant. * WER and many other metrics poorly
Medical ontologies are like maps of diseases, drugs, and symptoms. They power search, decision support, and research—but they miss many real-world terms used in doctors’ notes. A new framework called CLOZE taps large language models to read clinical notes and suggest new concepts and where they fit in the
AI
The debate over AI consciousness is loud—and often confusing. This paper offers a simple map to sort arguments, so we can see where people disagree and how strongly they mean it. What does an objection target? * The system’s goals and capabilities (what it does) — the computational level. * The
AI
Building capable embodied AI is hard: real-world data is scarce and training is expensive. This paper introduces Deliberate Practice Policy Optimization (DPPO), a coach-like training loop that helps vision-language agents learn more from less. How it works: the system alternates between learning from examples (to expand skills) and trial-and-error reinforcement
Medical AI is powerful, but it can sound more certain than it should. MedBayes-Lite is a lightweight add-on that helps transformer models say "I'm unsure" when cases are ambiguous, with no retraining and under 3% extra parameters. * Calibrated embeddings: Monte Carlo dropout captures what the model
Tiny synthetic images, big model results Training today’s vision systems usually starts from large, self-supervised backbones. This work shows you can replace huge real datasets with a handful of smartly crafted, synthetic images—without losing performance. The trick, called Linear Gradient Matching, distills a small set of images so
What if robots could reflect on their own decisions—like asking, "How sure am I?" A new study proposes a metacognition architecture that lets robots estimate confidence in their choices and use it to guide action. Inspired by neuroscience, the system monitors embodied actions in real time and
AI
Wearables are reshaping healthcare, but most AI models track one physical condition at a time and overlook depression—despite frequent overlap. This study reframes multi-disease assessment as a multi-task learning problem, letting one model learn several conditions jointly and use their relationships to improve accuracy. The catch is “double heterogeneity”
AI
Words like "bank" can mean a riverbank or a financial institution. Getting the right meaning—called word sense disambiguation—is hard for AI, especially when we need rich, logic-friendly knowledge for reasoning. This paper introduces an annotation-free workflow that teams up symbolic language understanding with large language models:
What if a single photo could become a detailed 3D object—geometry, texture, and where it sits in the scene? Meet SAM 3D, a generative model for visually grounded 3D reconstruction in the wild. * One image in, 3D out: Rebuilds shape, texture, and layout from everyday photos with occlusion and
Robotics
Robots are great at repeating tasks—but not at knowing when to trust their own choices. This study brings metacognition (self-monitoring) to robots by giving them a sense of confidence in each decision. * Inspired by neuroscience, the architecture treats confidence as a second-order judgment about actions. * In tests on autonomous
AI
Wearables and AI are reshaping healthcare, but most tools track only physical diseases. That misses depression, which often coexists with chronic conditions. This study introduces one model that assesses multiple chronic diseases and depression together, using data from everyday sensors. It treats each disease as its own task, so the