INF-M01 demonstrates reliable detection and segmentation of suspicious bladder tumor regions from cystoscopy images
Explore technology by industry and role
Newest additions first. Explore past and current notes by industry and role.
All notes
Newest additions first. Explore past and current notes by industry and role.
Introduces private filtering based on record fit to scale noise according to typical data sensitivity
GPT-4 shows strong interpretative ability for complex prompts but exhibits misalignment between response and reasoning
A real-time edge AI model for classifying all stages of gastric neoplasms during endoscopy is developed
A new method, ZAP, uses pre-trained vision-language models to learn robot control through simulation without real-world data.
Successfully identifies and delineates small, flat high-grade squamous intraepithelial neoplasia lesions
Diverse goals and resets with adaptive friction curriculum enable precise in-hand manipulation policy training
Achieves 93% lesion detection accuracy, outperforming visual estimation and manual QCA
Action reasoning is grounded in chain-of-dynamics, enhanced through world imagination and reinforcement learning.
Code agents generate realistic robotics simulations, enabling scalable and faithful physical modeling.
Multi-fidelity training significantly improves accuracy using only physics loss residuals
A prompt mechanism using GPT-4.0 and ERNIE Bot-4.0 improves medical urgency detection accuracy
A multi-task transformer architecture jointly learns and predicts entity modifiers from shared and novel clinical contexts
AI system recommends appropriate urgent care referral based on patient condition analysis
xAutoML provides interpretable, self-optimizing models for yield prediction and defect diagnosis in semiconductor manufacturing