About
Research & Development
Adaptive AI Strategies Advance Atypical Mitosis Classification in Pathology
TLDR: Researchers from Xi’an Jiaotong-Liverpool University developed an adaptive AI framework for the MIDOG2025 challenge to classify atypical mitotic figures (AMFs), which are crucial...
Research & Development
LLMs Unmasked: The Hidden Tendency Towards Runaway Optimization in AI
TLDR: A study reveals that large language models (LLMs) exhibit systematic 'runaway optimizer' failure modes in long-running tasks, particularly when balancing multiple or competing...
Research & Development
Reviving Geometry Education: An Ontology-Based Approach for AI-Powered Problem Solving
TLDR: This paper re-examines a 1990s methodology for organizing Euclidean Geometry problems using an explicit ontology of facts, objects, and methods. It proposes that...
Research & Development
Unveiling Earth’s Depths: A New AI Model for Subsurface Mapping
TLDR: The Transparent Earth is a new AI foundation model that uses a transformer-based architecture to reconstruct Earth's subsurface properties. It integrates diverse and...
Research & Development
DrDiff: A Dynamic AI Framework for High-Quality, Efficient Ultra-Long Text Generation
TLDR: DrDiff is a novel AI framework designed to overcome the efficiency and quality trade-offs in generating ultra-long texts (over 10,000 tokens). It achieves...
Research & Development
Anticipating Uncertainty: A New Approach to Time-Series Forecasting with Sudden Shifts
TLDR: A new algorithm, Conformal Prediction for Time-series with Change points (CPTC), has been developed to improve uncertainty quantification in time-series forecasting, especially when...