Archived Intelligence
Researchers have developed a hybrid artificial intelligence model combining Convolutional Neural Networks and Vision Transformers to identify crop diseases. Early tests on specific datasets suggest the system achieves high accuracy by mathematically focusing on infected plant tissue rather than background noise.
Deep learning models often fail to predict chemical behaviour when training data is limited. A new preprint suggests an AI model called CheMeleon can solve this by learning from clean, fundamental chemical rules rather than messy experimental data.
A new phylogenetic study traces the evolutionary history of marine light sensors to an ancient, non-photosynthetic ancestor. These findings challenge previous biological assumptions about how marine organisms adapted to extreme polar environments.
Researchers have developed a mathematical model to predict the success of brain surgery in Parkinson's patients. By mapping the exact overlap between electrical fields and a neural 'sweetspot', the study suggests outcomes can be reliably forecasted.
An early-stage study details a new automated system that translates flat 2D garment designs into highly accurate 3D models. By digitising the prototyping phase, this technology could drastically reduce production times and design errors for complex clothing.
Physicists have discovered a way to electrically control how a single iron atom interacts with particles inside a quantum dot. This precise tuning could lead to highly advanced single-atom electronics and quantum processors.
A new preprint study suggests a financial framework to turn mining emissions into a lucrative asset for climate bond investors. By factoring avoided emissions into project valuations, emerging economies could fund the shift to renewable-powered extraction.
Researchers have developed a new chemical ligand that significantly improves the durability and efficiency of next-generation solar materials. By fixing structural defects at the molecular level, the cells achieved over 26 percent efficiency and maintained most of their power after 1,000 hours of rigorous testing.
A new early-stage study reveals that popular biological age trackers measure distinct, heterogeneous cellular processes. By pairing DNA methylation data with RNA sequencing, researchers created new gene scores that may better predict age-related disease.
A new study reveals that visceral fat may quietly sabotage how the body processes vitamin D in patients undergoing haemodialysis. By tracking hormonal shifts over a year, researchers found that deep belly fat increases leptin and FGF-23, potentially choking off the active form of the vitamin.
A new computational pipeline uses machine learning and deep learning to identify uncharacterised protein targets in Chlamydia pneumoniae. While still early-stage and awaiting peer review, this method points toward a faster, data-driven future for antimicrobial research.
Traditional forest assessment is too slow for modern ecological repair. A new systematic review suggests that advanced AI models can identify tree species with over 96% accuracy, offering a swift new tool for ecologists.