首页|Study Findings on Artificial Intelligence Detailed by Researchers at Physikalisc h-Technische Bundesanstalt (Benchmarking the influence of pre-training on explan ation performance in MR image classification)
Study Findings on Artificial Intelligence Detailed by Researchers at Physikalisc h-Technische Bundesanstalt (Benchmarking the influence of pre-training on explan ation performance in MR image classification)
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 publish new report on ar tificial intelligence. According to news originating from Berlin, Germany, by Ne wsRx correspondents, research stated, "Convolutional Neural Networks (CNNs) are frequently and successfully used in medical prediction tasks." Funders for this research include European Research Council; Bundesministerium F ur Wirtschaft Und Klimaschutz. Our news editors obtained a quote from the research from Physikalisch-Technische Bundesanstalt: "They are often used in combination with transfer learning, lead ing to improved performance when training data for the task are scarce. The resu lting models are highly complex and typically do not provide any insight into th eir predictive mechanisms, motivating the field of ‘explainable' artificial inte lligence (XAI). However, previous studies have rarely quantitatively evaluated t he ‘explanation performance' of XAI methods against ground-truth data, and trans fer learning and its influence on objective measures of explanation performance has not been investigated. Here, we propose a benchmark dataset that allows for quantifying explanation performance in a realistic magnetic resonance imaging (M RI) classification task. We employ this benchmark to understand the influence of transfer learning on the quality of explanations. Experimental results show tha t popular XAI methods applied to the same underlying model differ vastly in perf ormance, even when considering only correctly classified examples."
NETL
NSTL
