人工智能培训

搜索

深度学习论文:使用超进程模型的深度绘图过程中的零镜头学习应用程序(A Zero-Shot Learning application in Deep Drawin

[复制链接]
kkkgame 发表于 2019-1-28 11:40:24 | 显示全部楼层 |阅读模式
kkkgame 2019-1-28 11:40:24 127 0 显示全部楼层
深度学习论文:使用超进程模型的深度绘图过程中的零镜头学习应用程序(A Zero-Shot Learning application in Deep Drawing process using  Hyper-Process Model)在当今工业化的世界中,从大规模生产转向大规模定制范式的后果之一是需要提高制造公司的灵活性和响应能力。高混合/低产量生产迫使未知产品变量的恒定调节,这最终导致机器校准的高周期。与机器校准相关的困难在于需要经验以及一组实验以满足最终产品质量。遗憾的是,机器参数的所有可能组合都很高,难以建立经验知识。由于这一事实,通常采用试验性的反恐方法,使得独一无二的产品不可行。因此,基于零镜头学习(ZSL)的超级过程模型(HPM)方法容许多个任务之间的关系被用作缩短校准阶段的方法。假设每个产品变体是要解决的任务,首先,对数据进行形状分析以学习任务之间的共同变形模式,其次,执行这些模式和任务描述之间的映射。最终,目前的工作有两个主要贡献:1)将工业问题制定为ZSL设置,其中可以生成用于过程优化的新过程模型和2)ZSL域中的入侵问题的定义。为此,基于从Abaqus模拟器收集的数据使用了二维深度图纸模拟过程,其中收集了大量过程模型以测试该方法的有效性。获得的结果表明,通过平均有关已有任务的信息,可以在没有任何可用数据(标记和未标记)的情况下容忍新任务,从而加快校准阶段并使新产品更快地集成到制造系统中。
One of the consequences of passing from mass production to mass customizationparadigm in the nowadays industrialized world is the need to increaseflexibility and responsiveness of manufacturing companies.The high-mix /low-volume production forces constant accommodations of unknown productvariants, which ultimately leads to high periods of machine calibration.Thedifficulty related with machine calibration is that experience is requiredtogether with a set of experiments to meet the final product quality.Unfortunately, all possible combinations of machine parameters is so high thatis difficult to build empirical knowledge.Due to this fact, normally trial anderror approaches are taken making one-of-a-kind products not viable.Therefore,a Zero-Shot Learning (ZSL) based approach called hyper-process model (HPM) tolearn the relation among multiple tasks is used as a way to shorten thecalibration phase.Assuming each product variant is a task to solve, first, ashape analysis on data to learn common modes of deformation between tasks ismade, and secondly, a mapping between these modes and task descriptions isperformed.Ultimately, the present work has two main contributions: 1)Formulation of an industrial problem into a ZSL setting where new processmodels can be generated for process optimization and 2) the definition of aregression problem in the domain of ZSL.For that purpose, a 2-d deep drawingsimulated process was used based on data collected from the Abaqus simulator,where a significant number of process models were collected to test theeffectiveness of the approach.The obtained results show that is possible tolearn new tasks without any available data (both labeled and unlabeled) byleveraging information about already existing tasks, allowing to speed up thecalibration phase and make a quicker integration of new products intomanufacturing systems.深度学习论文:使用超进程模型的深度绘图过程中的零镜头学习应用程序(A Zero-Shot Learning application in Deep Drawing process using  Hyper-Process Model) FKZKZKIkYe8ebbfE.jpg
URL地址:https://arxiv.org/abs/1901.08969     ----pdf下载地址:https://arxiv.org/pdf/1901.08969    ----深度学习论文:使用超进程模型的深度绘图过程中的零镜头学习应用程序(A Zero-Shot Learning application in Deep Drawing process using  Hyper-Process Model)
回复

使用道具 举报

您需要登录后才可以回帖 登录 | 立即注册

本版积分规则 返回列表 发新帖

kkkgame当前离线
新手上路

查看:127 | 回复:0

快速回复 返回顶部 返回列表