OpenConstruction: A Systematic Synthesis of Open Visual Datasets for Data-Centric Artificial Intelligence in Construction Monitoring

A recent study compares Continual Learning (CL) and Transfer Learning (TL) for modeling building thermal dynamics, particularly under changing conditions such as retrofits or occupancy changes. TL is highlighted as the most effective method when data is limited, utilizing pretrained models that can be fine-tuned with new operational data over time.

A recent study published in Nature — Machine Learning introduces a bio-friendly and high-precision super-resolution imaging technique through self-supervised reconstruction structured illumination microscopy. This innovative approach aims to enhance imaging capabilities while minimizing environmental impact, marking a significant advancement in the field of imaging technology.

GlyContact has developed a method for analyzing glycan 3D structures at scale, utilizing advanced machine learning techniques to enhance the understanding of glycan interactions and their biological significance. This innovation represents a significant step forward in glycomics research, enabling more comprehensive studies of glycans in various biological contexts.

A recent study published in Nature — Machine Learning demonstrates that knowledge-guided adaptation of pathology foundation models significantly enhances cross-domain generalization and demographic fairness in medical diagnostics. This advancement is crucial for improving the accuracy of pathology assessments across diverse populations.

A new machine learning model named GerNA-Bind has been developed to decipher RNA-ligand binding specificity, as reported in Nature — Machine Learning. This model aims to enhance the understanding of how RNA interacts with various ligands, which is crucial for advancing research in molecular biology and drug discovery.

A recent study published in Nature — Machine Learning has introduced an optimized MobileNet model designed for lightweight and accurate detection of plant leaf diseases. This advancement leverages machine learning techniques to improve agricultural diagnostics, enabling faster and more efficient identification of plant health issues.

FIDDLE, a deep learning method for predicting chemical formulas from tandem mass spectra, has been introduced in a recent publication by Nature — Machine Learning. This innovative approach aims to enhance the accuracy of chemical formula predictions, which is crucial for various applications in chemistry and related fields.

A novel methodology has been introduced for generating large-scale defect datasets in optical lithography, addressing the critical shortage of high-quality training data for artificial intelligence applications in the semiconductor industry. This approach utilizes physics-constrained mathematical morphology operations to synthesize defect layouts, which are then fabricated into physical samples for analysis.