在许多制造场景中,为机器学习和计算机视觉标注数据的成本很高,但合成数据的生成成本却低得多。因此,对于需要大量训练数据的机器学习应用而言,用合成数据替代真实数据极具吸引力。然而,仅依赖合成数据通常不足以有效训练在真实数据上表现良好的模型,这主要是由于合成数据与真实数据之间存在领域偏移。我们探讨了在检测铝制车轮的 X 射线扫描中的缺陷时,处理此类领域偏移的方法。我们使用模拟和真实的 X 射线图像,采用不同策略训练目标检测模型,以确定在最小化对标注真实训练样本需求的同时,能产生最佳检测结果的训练方法。我们的初步发现表明,如果可用标注样本总数固定,sim-2-real 领域自适应方法比完全监督的 oracle 模型更具成本效益。给定一定数量的已标注真实样本,在合成数据与未标注真实数据的混合集上进行训练,能以显著更低的成本取得相当甚至更好的检测结果。我们认为,未来对不同训练策略成本效益的研究对于更好地理解如何在实际机器学习项目中分配预算具有重要意义。
英文摘要:
In many manufacturing settings, annotating data for machine learning and computer vision is costly, but synthetic data can be generated at significantly lower cost. Substituting the real-world data with synthetic data is therefore appealing for many machine learning applications that require large amounts of training data. However, relying solely on synthetic data is frequently inadequate for effectively training models that perform well on real-world data, primarily due to domain shifts between the synthetic and real-world data. We discuss approaches for dealing with such a domain shift when detecting defects in X-ray scans of aluminium wheels. Using both simulated and real-world X-ray images, we train an object detection model with different strategies to identify the training approach that generates the best detection results while minimising the demand for annotated real-world training samples. Our preliminary findings suggest that the sim-2-real domain adaptation approach is more cost-efficient than a fully supervised oracle - if the total number of available annotated samples is fixed. Given a certain number of labeled real-world samples, training on a mix of synthetic and unlabeled real-world data achieved comparable or even better detection results at significantly lower cost. We argue that future research into the cost-efficiency of different training strategies is important for a better understanding of how to allocate budget in applied machine learning projects.
AI-Guided Defect Detection Techniques to Model Single Crystal Diamond Growth
Fraunhofer USA, Center Mid-Atlantic(弗劳恩霍夫美国中心中大西洋分部);Michigan State University(密歇根州立大学);Fraunhofer USA, Center Midwest(弗劳恩霍夫美国中心中西部部分);University of Maryland(马里兰大学)
Hard to See, Hard to Label: Generative and Symbolic Acquisition for Subtle Visual Phenomena
University of South Carolina(南卡罗来纳大学);AI Research Lab, HPE Labs, Hewlett Packard Enterprise(HPE实验室人工智能研究部);Indian AI Research Organization(印度人工智能研究组织);General Atomics(通用原子公司)
Automated Defect Detection for Mass-Produced Electronic Components Based on YOLO Object Detection Models
Department of Electrical Engineering and Graduate School of Engineering Science and Technology(电气工程系和工程科学与技术研究生学院);National Yunlin University of Science and Technology(云林科技大学);Research Center for Information Technology Innovation(信息技术创新研究中心);Academia Sinica(学术院);Institute of Communications Engineering(通讯工程研究所);National Sun Yat-sen University(国立中山大学)
Segmentation in large-scale cellular electron microscopy with deep learning: A literature survey
Bernoulli Institute of Mathematics, Computer Science and Artificial Intelligence, University Groningen(格罗宁根大学伯努利数学、计算机科学与人工智能研究所);University Groningen(格罗宁根大学);University Medical Center Groningen(格罗宁根大学医学中心)
Active Transfer Bagging: A New Approach for Accelerated Active Learning Acquisition of Data by Combined Transfer Learning and Bagging Based Models
Chemical Engineering School for the Engineering of Matter, Transport, & Energy Arizona State University(材料与运输工程学院,亚利桑那州立大学);Materials Science & Engineering School for the Engineering of Matter, Transport, & Energy Arizona State University(材料科学与工程学院,亚利桑那州立大学)
Towards a Pseudo-Labeling Workflow for Celltype-Classification from Explanted Brain Slice Recordings
RWTH University Hospital, Department of Neurosurgery(RWTH大学医院神经外科部门);RWTH University Hospital, Department of Epileptology(RWTH大学医院癫痫学部门);University of Duisburg-Essen, Intelligent Embedded Systems Lab(杜伊斯堡-埃森大学智能嵌入式系统实验室)
Active Learning Using Aggregated Acquisition Functions: Accuracy and Sustainability Analysis
Chair of Information Theory and Data Analytics (INDA), RWTH Aachen University(信息理论与数据 analytics 教授会(INDA),亚琛工业大学);Automation and Control Institute, Technische Universität Wien (TUW)(自动化与控制研究所,维也纳技术大学);AIT Austrian Institute of Technology GmbH(奥地利技术研究所(AIT))
Robust Active Learning: Sample-Efficient Training of Robust Deep Learning Models
University of Luxembourg(卢森堡大学)
主动学习是一种已建立的技术,用于降低标记成本以构建高质量的机器学习模型。主动学习的一个核心组成部分是获取功能,它确定应该选择哪些数据进行注释。最先进的采集功能--以及更大程度上的主动学习技术--被设计成最大化干净性能(例如准确性),而忽略了鲁棒性,鲁棒性是一个越来越受到关注的重要质量属性。因此,主动学习产生的模型是准确的,但并不稳健。在本文中,我们提出了\emph{robust active learning},这是一个集成了对抗训练的主动学习过程--这是产生鲁棒模型的最成熟的方法。通过对11个采集函数,4个数据集,6个DNN架构和15105个训练DNN的实证研究,我们表明鲁棒主动学习可以产生鲁棒性(对抗性示例的准确性)范围从2.35%到63.85%的模型,而标准主动学习系统地实现了可忽略的鲁棒性(小于0.20%)。然而,我们的研究还表明,在准确性方面表现良好的采集函数在鲁棒性方面比随机采样差。因此,我们研究这背后的原因,并设计了一个新的采集功能,目标是干净的性能和鲁棒性。我们的采集函数-命名为基于密度的鲁棒采样熵(DRE)-优于其他采集函数(包括随机)的鲁棒性高达24.40%(3.84%,特别是比随机),同时保持竞争力的准确性。此外,我们证明了DRE是适用于作为模型再训练的测试选择度量,并从所有比较的功能中脱颖而出,高达8.21%的鲁棒性。
Towards Computationally Feasible Deep Active Learning
AIRI(人工智能研究所(AIRI));FRC CSC RAS(俄罗斯科学院计算机科学与控制联邦研究中心);MIPT(莫斯科物理技术学院);HSE(高等经济学院);Sber AI Lab(俄罗斯联邦储蓄银行人工智能实验室);ISP RAS Research Center for Trusted Artificial Intelligence(俄罗斯科学院系统编程研究所可信人工智能研究中心)
Active Learning for Automated Visual Inspection of Manufactured Products
University of Ljubljana(卢布尔雅那大学);Jožef Stefan International Postgraduate School(约热夫·斯特凡国际研究生院);Philips Consumer Lifestyle BV(飞利浦消费生活方式有限公司);Qlector d.o.o.(Qlector有限责任公司);Jožef Stefan Institute(约热夫·斯特凡研究所)