首页|New Machine Learning Study Findings Have Been Reported from University College L ondon (UCL) (Tackling Data Scarcity Challenge Through Active Learning In Materia ls Processing With Electrospray)
New Machine Learning Study Findings Have Been Reported from University College L ondon (UCL) (Tackling Data Scarcity Challenge Through Active Learning In Materia ls Processing With Electrospray)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News-Investigators discuss new findings in Machine Learning. According to news originating from London, United Kingdom, by NewsRx correspondents, research stated, "Machine learning (ML) has been harnesse d as a promising modelling tool for materials research. However, small data, or data scarcity, is a bottleneck when incorporating ML in studies involving experi mentation." Financial support for this research came from Engineering and Physical Sciences Research Council. Our news journalists obtained a quote from the research from University College London (UCL), "Current experiment planning methods show several disadvantages: o ne-factor-at-a-time (OFAT) experimentation became impractical due to limited lab oratory resources; conventional design of experiments (DoE) failed to incorporat e high-dimensional features in ML; Surrogate-based or Bayesian optimization (BO) shifted the goal to optimize material properties rather than guiding training d ata accumulation. The present research proposes leveraging active learning (AL) to strategically select critical data for experimentation. Two AL strategies, qu ery-by-Committee (QBC) algorithm and Greedy method, are benchmarked against rand om query baseline on various materials datasets. AL is shown to efficiently redu ce model prediction errors with minimal additional experiment data. Investigatio n of hyperparameters revealed benefits of applying AL at an early stage of exper imental dataset construction. Moreover, AL is implemented and validated for an i n-house materials development task - electrospray modelling. AL exploration as a paradigm is highlighted to guide experiment design for efficient data accumulat ion purposes, and its potential for further ML modelling. In doing so, the power of ML is expected to be fully unleashed to experimental researchers. Small data is a prevalent bottleneck in machine learning for materials research. This stud y suggests active learning (AL) as a new paradigm for data acquisition. Through strategical selection, AL recommends information-rich datapoints for laboratory investigation."
LondonUnited KingdomEuropeCyborgsEmerging TechnologiesMachine LearningUniversity College London (UCL)
NETL
NSTL
