Single machine scheduling with a general exponential learning effect

In this paper we consider the scheduling problem with a general exponential learning effect and past-sequence-dependent (p-s-d) setup times. By the general exponential learning effect, we mean that the processing time of a job is defined by an exponent function of the total weighted normal processing time of the already processed jobs and its position in a sequence, where the weight is a position-dependent weight. The setup times are proportional to the length of the already processed jobs. We consider the following objective functions: the makespan, the total completion time, the sum of the δ ? 0th power of completion times, the total weighted completion time and the maximum lateness. We show that the makespan minimization problem, the total completion time minimization problem and the sum of the quadratic job completion times minimization problem can be solved by the smallest (normal) processing time first (SPT) rule, respectively. We also show that the total weighted completion time minimization problem and the maximum lateness minimization problem can be solved in polynomial time under certain conditions.     

A new comparison theorem on conditional quantiles

This paper studies the conditional quantile regression problem involving the pinball loss. We introduce a concept of τ-quantile of p-average logarithmic type q to complement the previous study by Steinwart and Christman (2008, 2011) [1] and [2]. A new comparison theorem is provided which can be used for further error analysis of some learning algorithms.     

Inventory models for imperfect quality items with shortages and learning in inspection

The classical economic order quantity model, although well known and useful; assumes that all items received conform to quality characteristics. However, in practice, items may be damaged due to transportation and/or production conditions. This requires a buyer to screen each lot it receives from its vendor to separate the good from the nonconforming (due to imperfect quality) items. While screening is usually a manual task performed by inspectors, it may improve with learning. Besides, it was observed in some studies that coordinating activities (e.g., quality) between a buyer and a vendor may be subject to learning effects and results in improving the quality of each lot (as it contains less nonconforming items) delivered or produced.     

基于KINGOSOFT平台的政治经济学教学模式初探

基于Kingosoft(课程管理系统(CMS))平台,结合大学政治经济学的教学实际,探讨了在Kingosoft平台中的教学模式,为广大教学设计者在Kingosoft平合中更好地设计教学活动提供了一定的理论依据.     

基于光谱技术的双孢蘑菇新鲜度量化检测技术研究

双孢蘑菇质地柔嫩、营养丰富,具有很好的降血压、降血脂、消炎护肝等多种保健价值,其新鲜度是反映内外部品质的重要指标之一。目前双孢蘑菇新鲜度鉴别大多依据其外观品质变化（褐变）,缺乏精准的量化评价指标与方法,因此提出了以贮藏天数为新鲜度检测的量化指标,并利用近红外光谱技术对双孢蘑菇新鲜度进行检测分析。依据存储天数不同,将双孢蘑菇样本分为1~5组,每组40个样本,依次采集每组双孢蘑菇的近红外光谱数据。针对采集的原始光谱数据,首先选用卷积平滑滤波（SG）与多元散射校正（MSC）消除原始光谱噪声、基线平移以及光散射的影响,并选取399.81~999.81 nm的光谱波段作为数据处理范围;然后分别使用主成分分析（PCA）和连续投影算法（SPA）进行光谱降维和特征波长选择,继而建立极限学习机（ELM）分类模型;同时考虑到ELM模型中初始值对分类准确率影响较大,分别选用粒子群优化算法（PSO）、海鸥优化算法（SOA）对ELM中初始权值及阈值进行寻优,形成PSO-ELM,SOA-ELM优化组合分类模型;最后分别将全光谱、提取主成分以及所选的特征波长｛556.87,445.51,481.15,885.10,802.25,720.90,861.34,909.79,924.44,873.17 nm｝输入到分类模型中,建立不同输入、不同分类模型的双孢菇新鲜度检测模型。最终试验结果表明,当ELM为分类模型,以全光谱、主成分以及特征波长为输入时的预测精度分别为75%,95%,88%;以SPA优选特征波长作为输入的PSO-ELM、SOA-ELM分类模型训练集精度为96.25%,93.25%,预测集精度为92.5%,94%。可知,SPA波长选择算法可以有效降低光谱信息中存在的冗余信息,加快建模效率,同时海鸥优化算法能较好的优化ELM分类模型的初始参数,分类精度较ELM模型提高了6.8%,同时不产生过拟合现象。因此,利用光谱特征可以快速、准确无损的识别双孢蘑菇的新鲜度,研究结果为便携式双孢蘑菇新鲜度快速无损检测设备的开发提供了理论依据。     

高光谱漫反射无损检测鸡蛋受精状态

种鸡蛋孵化期间受精状态的检测需要消耗大量人力、物力,并且孵化期间的种鸡蛋不能保证均为健康蛋,需要能够在孵化早期将无精蛋和死精蛋快速准确挑选出来达到降低生产成本的目的。以白来航鸡蛋为研究对象,采用高光谱分选仪批量采集受精、未受精、死精三类鸡蛋共119枚在382～1 026 nm范围内的高光谱数据,其中受精蛋采集孵化3, 5, 7, 9, 11, 13和15 d的数据,并通过黑白校正方法对原始光谱图做校正处理,得到其漫反射率,经过实验对比以及根据实际生产需要,受精蛋选用孵化3和5 d的光谱数据作为建模数据。同时提出了一种将光谱数据转换为图像数据的方法,在最大化保证光谱原始数据的前提下达到了光谱向量数据可视化的效果,可以有效与深度学习图像识别算法相结合。采用连续投影算法(SPA)、竞争性自适应重加权算法(CARS)对光谱波段进行筛选,建立基于全波段、 CARS筛选的特征波长、 SPA筛选的特征波长与SVM、 RandomForest算法与AlexNet、 MobileNet网络的判别模型,其中AlexNet-5dFull Wave Bands准确率最高为93.22%。与通过不同特征波长算法筛...     

Information Bottleneck Theory Based Exploration of Cascade Learning

In solving challenging pattern recognition problems, deep neural networks have shown excellent performance by forming powerful mappings between inputs and targets, learning representations (features) and making subsequent predictions. A recent tool to help understand how representations are formed is based on observing the dynamics of learning on an information plane using mutual information, linking the input to the representation (I(X;T)) and the representation to the target (I(T;Y)). In this paper, we use an information theoretical approach to understand how Cascade Learning (CL), a method to train deep neural networks layer-by-layer, learns representations, as CL has shown comparable results while saving computation and memory costs. We observe that performance is not linked to information–compression, which differs from observation on End-to-End (E2E) learning. Additionally, CL can inherit information about targets, and gradually specialise extracted features layer-by-layer. We evaluate this effect by proposing an information transition ratio, I(T;Y)/I(X;T), and show that it can serve as a useful heuristic in setting the depth of a neural network that achieves satisfactory accuracy of classification.     

A two-level Iterative Learning Control scheme for the engagement of wet clutches

This paper discusses the application of Iterative Learning Control (ILC) algorithms for the engagement of wet clutches. A two-level control scheme is presented, consisting of a high level ILC-type algorithm which iteratively updates parameterized reference trajectories which are tracked by the low level tracking control. At this low level, two standard ILC controllers are used to first track a pressure reference in the filling phase and afterwards a slip reference in the slip phase of the clutch engagement. The performance and robustness of the presented approach are validated on an experimental test setup. It is shown that both levels are crucial to achieve good engagement quality during normal machine operation. Through the use of this ILC control scheme, it is possible to avoid time-consuming and cumbersome experimental (re)calibrations, which are nowadays used to achieve and maintain good performance despite the complex and time-varying dynamics of wet clutches.     

Towards an integrated theory of mathematics conceptual learning and instructional design: The Learning Through Activity theoretical framework

We discuss the theoretical framework of the Learning Through Activity research program. The framework includes an elaboration of the construct of mathematical concept, an elaboration of Piaget’s reflective abstraction for the purpose of mathematics pedagogy, further development of a distinction between two stages of conceptual learning, and a typology of different reverse concepts. The framework also involves instructional design principles built on those constructs, including steps for the design of task sequences, development of guided reinvention, and ways of promoting reversibility of concepts. This article represents both a synthesis of prior work and additions to it.     

Denoising Convolutional Neural Network with Energy-Based Attention for Image Enhancement

In the realm of image denoising, the use of convolutional neural networks (CNNs) has lately gained traction. Several activities involve the utilization of excellent-clarity pictures and recordings. Images were captured in a wide variety of illumination circumstances, which means that not all of them are of the highest quality. Low-light photography suffers from a decline in perceived image quality because of the restricted dynamic range of the pixel values. Therefore, it is vital to enhance the appearance of images. Maximum texture retention is achieved by the structural similarity index-loss-based method. The suggested discrete wavelet transform (DWT)-self attention (SA)-Denoising convolutional neural networks (DnCNNs) make use of state-of-the-art techniques for image denoising like energy band analysis, very deep architecture, learning algorithms, dense-sparse-dense training, and regularization approaches. DnCNN is intended to remove the hidden layers" latent, yielding a pure picture. After a degraded input sample has had its relevant energy features retrieved using DWT, the perfect image enhancement is achieved thanks to the incorporation of the self-attention mechanism. Second, a hierarchical-branch network is formed by combining the suggested network with the denoising CNN and additional loss in order to reduce the reliance on the amount of noisy data in multi-modal picture analysis and make the problem of image enhancement more tractable. In the end, DWT-SA-DnCNN"s self-learning qualities are used to improve image quality by obtaining features including undesirable noisy data, edge factor, texture, uniform and non-uniform areas, smoothness, and object structure. Simulation results show that our hybrid DWT-SA-DnCNN-based contrast enhancement strategy outperforms state-of-the-art methods.