IFFair: Influence Function-driven Sample Reweighting for Fair Classification

Researchers at Edith Cowan University are leveraging artificial intelligence (AI) and machine learning to enhance the reliability of solid-state batteries, addressing a significant challenge in battery technology. This initiative aims to improve performance and safety in energy storage solutions.

A new paper has been published on arXiv detailing an unsupervised learning approach for density estimation using a topology-based loss function. This method aims to automate the selection of the optimal kernel bandwidth, a critical hyperparameter that influences the bias-variance trade-off in density estimation, particularly in high-dimensional data where visualization is challenging.

A new automated machine learning framework has been developed to profile the morphology of 316L powder particles for Selective Laser Melting (SLM) in additive manufacturing. This approach utilizes high-throughput imaging, shape extraction, and clustering to analyze approximately 126,000 powder images, significantly enhancing the characterization process compared to traditional methods.

A study has introduced a machine learning framework for predicting the California Bearing Ratio (CBR) using a dataset of 382 soil samples from various geoclimatic regions in Tükiye. This approach aims to enhance the accuracy and efficiency of CBR determination, which is crucial for assessing the load-bearing capacity of subgrade soils in infrastructure projects.

A recent study published in Nature — Machine Learning presents a machine learning model capable of predicting the immune age of mice based on cellular patterns. This innovative approach leverages complex data analysis to enhance understanding of immune system aging, potentially leading to advancements in immunology and age-related research.

The recent publication introduces an Alternating Preconditioned Gradient Descent (APGD) algorithm aimed at enhancing low-tubal-rank tensor estimation, a crucial task in high-dimensional signal processing and machine learning. Traditional methods, reliant on tensor singular value decomposition, are computationally intensive and impractical for large tensors, prompting the need for more efficient solutions.

A new concept called Interpretive Efficiency has been introduced, which quantifies how effectively data supports interpretive representations in machine learning. This measure is grounded in five axioms and relates to mutual information, providing a framework for assessing the usefulness of data in interpretive tasks.

Tyche introduces a novel approach to medical image segmentation by utilizing stochastic in-context learning, allowing for predictions on new tasks without retraining. This model addresses the limitations of existing methods that require extensive resources and expertise for each new segmentation task, and it acknowledges the inherent uncertainty in segmentation outcomes by generating multiple predictions.