Understanding is the essence of any intelligent system.We revise our four machine understanding paradigms which are:(i)basic understanding,(ii)rich understanding,(iii)exploratory understanding,and(iv)theory-based unde...Understanding is the essence of any intelligent system.We revise our four machine understanding paradigms which are:(i)basic understanding,(ii)rich understanding,(iii)exploratory understanding,and(iv)theory-based understanding;and we elaborate on the first two of them.We then introduce the concept of two-stage(or deep)machine understanding which provides descriptive understandings,as well as evaluations of these understandings.After a brief systematization of emotions,we cover the following paradigms for agents with two-stage(deep)understanding abilities for emotional intelligence simulation:(i)basic understanding,(ii)rich-understanding,and(iii)switchable understanding.展开更多
Prediction and control of chemical mixing are vital for many scientific areas such as subsurface reactive transport, climate modeling, combustion, epidemiology, and pharmacology. Due to the complex nature of mixing ...Prediction and control of chemical mixing are vital for many scientific areas such as subsurface reactive transport, climate modeling, combustion, epidemiology, and pharmacology. Due to the complex nature of mixing in heterogeneous andanisotropic media, the mathematical models related to this phenomenon are not analytically tractable. Numerical simulations often provide a viable route to predict chemical mixing accurately. However, contemporary modeling approaches for mixing cannot utilize available spatial-temporal data to improve the accuracy of the future prediction and can be compute-intensive, especially when the spatial domain is large andfor long-term temporal predictions. To address this knowledge gap, we will present inthis paper a deep learning (DL) modeling framework applied to predict the progress ofchemical mixing under fast bimolecular reactions. This framework uses convolutionalneural networks (CNN) for capturing spatial patterns and long short-term memory(LSTM) networks for forecasting temporal variations in mixing. By careful design ofthe framework—placement of non-negative constraint on the weights of the CNN andthe selection of activation function, the framework ensures non-negativity of the chemical species at all spatial points and for all times. Our DL-based framework is fast,accurate, and requires minimal data for training. The time needed to obtain a forecastusing the model is a fraction (≈ O(10−6)) of the time needed to obtain the result using a high-fidelity simulation. To achieve an error of 10% (measured using the infinitynorm) for capturing local-scale mixing features such as interfacial mixing, only 24%to 32% of the sequence data for model training is required. To achieve the same levelof accuracy for capturing global-scale mixing features, the sequence data required formodel training is 64% to 70% of the total spatial-temporal data. Hence, the proposedapproach—a fast and accurate way to forecast long-time spatial-temporal mixing patterns in heterogeneous and anisotropic media—will be a valuable tool for modelingreactive-transport in a wide range of applications.展开更多
文摘存储器是现代电子系统的核心器件之一,常用于满足不同层次的数据交换与存储需求.然而频率提高、时钟抖动、相位漂移以及不合理的布局布线等因素,都可能导致CPU对存储器访问稳定性的下降.针对同步动态随机读写存储器(synchronous dynamic random access memory,SDRAM)接口的时钟信号提出了一种自适应同步的训练方法,即利用可控延迟链使时钟相位按照训练模式偏移到最优相位,从而保证了存储器访问的稳定性.在芯片内部硬件上提供了一个可通过CPU控制的延迟电路,用来调整SDRAM时钟信号的相位.在系统软件上设计了训练程序,并通过与延迟电路的配合来达到自适应同步的目的:当CPU访问存储器连续多次发生错误时,系统抛出异常并自动进入训练模式.该模式令CPU在SDRAM中写入测试数据并读回,比对二者是否一致.根据测试数据比对结果,按训练模式调整延迟电路的延迟时间.经过若干次迭代,得到能正确访问存储器的延迟时间范围,即"有效数据采样窗口",取其中值即为SDRAM最优时钟相位偏移,完成训练后对系统复位,并采用新的时钟相位去访问存储器,从而保证读写的稳定性.仿真实验结果表明,本方法能迅速而准确地捕捉到有效数据采样窗口的两个端点位置,并以此计算出最佳的延迟单元数量,从而实现提高访问外部SDRAM存储器稳定性的目的.
文摘Understanding is the essence of any intelligent system.We revise our four machine understanding paradigms which are:(i)basic understanding,(ii)rich understanding,(iii)exploratory understanding,and(iv)theory-based understanding;and we elaborate on the first two of them.We then introduce the concept of two-stage(or deep)machine understanding which provides descriptive understandings,as well as evaluations of these understandings.After a brief systematization of emotions,we cover the following paradigms for agents with two-stage(deep)understanding abilities for emotional intelligence simulation:(i)basic understanding,(ii)rich-understanding,and(iii)switchable understanding.
文摘Prediction and control of chemical mixing are vital for many scientific areas such as subsurface reactive transport, climate modeling, combustion, epidemiology, and pharmacology. Due to the complex nature of mixing in heterogeneous andanisotropic media, the mathematical models related to this phenomenon are not analytically tractable. Numerical simulations often provide a viable route to predict chemical mixing accurately. However, contemporary modeling approaches for mixing cannot utilize available spatial-temporal data to improve the accuracy of the future prediction and can be compute-intensive, especially when the spatial domain is large andfor long-term temporal predictions. To address this knowledge gap, we will present inthis paper a deep learning (DL) modeling framework applied to predict the progress ofchemical mixing under fast bimolecular reactions. This framework uses convolutionalneural networks (CNN) for capturing spatial patterns and long short-term memory(LSTM) networks for forecasting temporal variations in mixing. By careful design ofthe framework—placement of non-negative constraint on the weights of the CNN andthe selection of activation function, the framework ensures non-negativity of the chemical species at all spatial points and for all times. Our DL-based framework is fast,accurate, and requires minimal data for training. The time needed to obtain a forecastusing the model is a fraction (≈ O(10−6)) of the time needed to obtain the result using a high-fidelity simulation. To achieve an error of 10% (measured using the infinitynorm) for capturing local-scale mixing features such as interfacial mixing, only 24%to 32% of the sequence data for model training is required. To achieve the same levelof accuracy for capturing global-scale mixing features, the sequence data required formodel training is 64% to 70% of the total spatial-temporal data. Hence, the proposedapproach—a fast and accurate way to forecast long-time spatial-temporal mixing patterns in heterogeneous and anisotropic media—will be a valuable tool for modelingreactive-transport in a wide range of applications.
