Modeling system could enable future generations of self-sensing materials

Phys.org  October 7, 2024 An international team of researchers (UK, Turkey) developed an experimentally informed predictive framework for autonomous sensing architected materials, combining theoretical and computational methodologies. The model incorporates stress-dependent electrical resistivity, geometric, and contact nonlinearities. It captures architecture-dependent piezoresistive responses of lattice composites produced via additive manufacturing of polyetherimide (PEI)/carbon nanotube (CNT) nanoengineered feedstock. The PEI/CNT composite exhibited exceptional strength, stiffness, and strain sensitivity, translating into remarkable piezoresistive characteristics for the PEI/CNT lattice composites, surpassing existing works. It accurately predicts both macroscopic piezoresistive responses and the influence of architectural and topological variations on electric current paths. According to […]

Study: Transparency is often lacking in datasets used to train large language models

MIT News  August 30, 2024 To improve data transparency and understanding of training language models on vast, diverse and inconsistently documented datasets an international team of researchers (USA – MIT, Harvard, UC Irvine, industry, University of Colorado, Olin College of Engineering, Carnegie Mellon University, and France, Canada) convened a multi-disciplinary effort between legal and machine learning experts to systematically audit and trace more than 1,800 text datasets. They developed tools and standards to trace the lineage of these datasets, including their source, creators, licenses and subsequent use. They found sharp divides in the composition and focus of data licensed for […]

How to figure out what you don’t know

TechXplore  October 26, 2020 Machine learning optimizes flexible models to predict data. In scientific applications, there is a rising interest in interpreting these flexible models to derive hypotheses from data. Researchers from Cold Spring Harbor Laboratory tested this connection using a flexible, yet intrinsically interpretable framework for modelling neural dynamics. Many models discovered during optimization predict data equally well, yet they fail to match the correct hypothesis. They developed an alternative approach that identifies models with correct interpretation by comparing model features across data samples to separate true features from noise. Their results reveal that good predictions cannot substitute for […]