基于延迟部分推理的快速前向规划系统

根据动作组件诱发关系的存在和抵制计算的必要性,提出一个计算松弛规划解的新方法——延迟部分推理.该方法在考虑动作删除效果的假定下,构造不包含任何互斥关系的组件规划图,通过定义"松弛诱发"关系预测后续规划过程中可能出现的组件诱发现象,在松弛规划解提取阶段判断动作组件间的"松弛诱发"关系并选择抵制动作避免可能发生的消极作用.基于延迟部分推理方法定义了新的启发式函数和剪枝策略,设计了规划系统FFc并在多个国际通用的测试域上进行实验.结果表明,FFc较之Fast-Forward在求解效率和求解质量方面都有显著的提高.     

Assessing of long-distance background nutrient loading to the Archipelago Sea, northern Baltic, with a hydrodynamic model

Nutrient concentrations have increased in the Archipelago Sea during recent years. This is probably caused by growth in the loads from agriculture, air, fish farming and the Baltic proper. The Archipelago Sea acts like a nutrient filter between the Baltic proper and the Bothnian Sea; part of some of the externally input nutrients are retained within this area. Previously only rough estimates of the amount of the background load have been given. In this study the amount of net nutrient inflow from the Baltic proper was estimated with the aid of a three-dimensional hydrodynamic model for the years 1993–1997. Our model suggests that the role of background nutrient load from the Baltic proper may sometimes be decisive in the eutrophication process. For example, in June 1993 the background load of phosphorus was 300 000 kg in one month, i.e. about threefold compared to the total phosphorus load from fish farming during this whole year. Nutrient inflow varies a lot during open water seasons and therefore more detailed studies are highly recommended. If all other local loading sources were included in the dynamic nutrient balance model we could calculate actual nutrient concentrations at any given time in the Archipelago Sea.     

引入概率分布的深度神经网络贪婪剪枝

目的 深度学习在自动驾驶环境感知中的应用,将极大提升感知系统的精度和可靠性,但是现有的深度学习神经网络模型因其计算量和存储资源的需求难以部署在计算资源有限的自动驾驶嵌入式平台上。因此为解决应用深度神经网络所需的庞大计算量与嵌入式平台有限的计算能力之间的矛盾,提出了一种基于权重的概率分布的贪婪网络剪枝方法,旨在减少网络模型中的冗余连接,提高模型的计算效率。方法 引入权重的概率分布,在训练过程中记录权重参数中较小值出现的概率。在剪枝阶段,依据训练过程中统计的权重概率分布进行增量剪枝和网络修复,改善了目前仅以权重大小为依据的剪枝策略。结果 经实验验证,在Cifar10数据集上,在各个剪枝率下本文方法相比动态网络剪枝策略的准确率更高。在ImageNet数据集上,此方法在较小精度损失的情况下,有效地将AlexNet、VGG（visual geometry group）16的参数数量分别压缩了5.9倍和11.4倍,且所需的训练迭代次数相对于动态网络剪枝策略更少。另外对于残差类型网络ResNet34和ResNet50也可以进行有效的压缩,其中对于ResNet50网络,在精度损失增加较小的情况下,相比目前最优的方法HRank实现了更大的压缩率（2.1倍）。结论 基于概率分布的贪婪剪枝策略解决了深度神经网络剪枝的不确定性问题,进一步提高了模型压缩后网络的稳定性,在实现压缩网络模型参数数量的同时保证了模型的准确率。     

Constraint satisfaction adaptive neural network and heuristicscombined approaches for generalized job-shop scheduling

This paper presents a constraint satisfaction adaptive neural network, together with several heuristics, to solve the generalized job-shop scheduling problem, one of NP-complete constraint satisfaction problems. The proposed neural network can be easily constructed and can adaptively adjust its weights of connections and biases of units based on the sequence and resource constraints of the job-shop scheduling problem during its processing. Several heuristics that can be combined with the neural network are also presented. In the combined approaches, the neural network is used to obtain feasible solutions, the heuristic algorithms are used to improve the performance of the neural network and the quality of the obtained solutions. Simulations have shown that the proposed neural network and its combined approaches are efficient with respect to the quality of solutions and the solving speed.     

Discriminant component pruning. Regularization and interpretation of multi-layered back-propagation networks

Neural networks are often employed as tools in classification tasks. The use of large networks increases the likelihood of the task's being learned, although it may also lead to increased complexity. Pruning is an effective way of reducing the complexity of large networks. We present discriminant components pruning (DCP), a method of pruning matrices of summed contributions between layers of a neural network. Attempting to interpret the underlying functions learned by the network can be aided by pruning the network. Generalization performance should be maintained at its optimal level following pruning. We demonstrate DCP's effectiveness at maintaining generalization performance, applicability to a wider range of problems, and the usefulness of such pruning for network interpretation. Possible enhancements are discussed for the identification of the optimal reduced rank and inclusion of nonlinear neural activation functions in the pruning algorithm.     

Adaptive selection of heuristics for assigning time slots and rooms in exam timetables

This paper presents an iterative adaptive approach which hybridises bin packing heuristics to assign exams to time slots and rooms. The approach combines a graph-colouring heuristic, to select an exam in every iteration, with bin-packing heuristics to automate the process of time slot and room allocation for exam timetabling problems. We start by analysing the quality of the solutions obtained by using one heuristic at a time. Depending on the individual performance of each heuristic, a random iterative hyper-heuristic is used to randomly hybridise the heuristics and produce a collection of heuristic sequences to construct solutions with different quality. Based on these sequences, we analyse the way in which the bin packing heuristics are automatically hybridised. It is observed that the performance of the heuristics used varies depending on the problem. Based on these observations, an iterative hybrid approach is developed to adaptively choose and hybridise the heuristics during solution construction. The overall aim here is to automate the heuristic design process, which draws upon an emerging research theme which is concerned with developing methods to design and adapt heuristics automatically. The approach is tested on the exam timetabling track of the second International Timetabling Competition, to evaluate its ability to generalise on instances with different features. The hyper-heuristic with low-level graph-colouring and bin-packing heuristics approach was found to generalise well over all the problem instances and performed comparably to the state of the art approaches.     

Research on an online self-organizing radial basis function neural network

A new growing and pruning algorithm is proposed for radial basis function (RBF) neural network structure design in this paper, which is named as self-organizing RBF (SORBF). The structure of the RBF neural network is introduced in this paper first, and then the growing and pruning algorithm is used to design the structure of the RBF neural network automatically. The growing and pruning approach is based on the radius of the receptive field of the RBF nodes. Meanwhile, the parameters adjusting algorithms are proposed for the whole RBF neural network. The performance of the proposed method is evaluated through functions approximation and dynamic system identification. Then, the method is used to capture the biochemical oxygen demand (BOD) concentration in a wastewater treatment system. Experimental results show that the proposed method is efficient for network structure optimization, and it achieves better performance than some of the existing algorithms.     

Order selection tests with multiply imputed data

Nonparametric tests for the null hypothesis that a function has a prescribed form are developed and applied to data sets with missing observations. Omnibus nonparametric tests such as the order selection tests, do not need to specify a particular alternative parametric form, and have power against a large range of alternatives. More specifically, likelihood-based order selection tests are defined that can be used for multiply imputed data when the data are missing-at-random. A simulation study and data analysis illustrate the performance of the tests. In addition, an Akaike information criterion for model selection is presented that can be used with multiply imputed datasets.     

A case-based approach to heuristic planning

Most of the great success of heuristic search as an approach to AI Planning is due to the right design of domain-independent heuristics. Although many heuristic planners perform reasonably well, the computational cost of computing the heuristic function in every search node is very high, causing the planner to scale poorly when increasing the size of the planning tasks. For tackling this problem, planners can incorporate additional domain-dependent heuristics in order to improve their performance. Learning-based planners try to automatically acquire these domain-dependent heuristics using previous solved problems. In this work, we present a case-based reasoning approach that learns abstracted state transitions that serve as domain control knowledge for improving the planning process. The recommendations from the retrieved cases are used as guidance for pruning or ordering nodes in different heuristic search algorithms applied to planning tasks. We show that the CBR guidance is appropriate for a considerable number of planning benchmarks.     

基于灵敏度分析的FPGM剪枝算法研究

针对等比例剪枝导致的重要卷积层剪枝过度、残留大量冗余参数以及精度损失较大的问题,在FPGM剪枝策略基础上融入灵敏度分析进行网络剪枝。算法采用精度反馈来分析每一层卷积层的重要性,控制单层剪枝比例分析每一层不同剪枝比例对精度损失的影响,获取各个卷积层的灵敏度;结合FPGM策略分析卷积层内卷积核的重要程度按灵敏度的剪枝比例剪掉不重要的卷积核,达到对神经网络进行有效剪枝的目的。实验结果表明,所提方法在MobileNet-v1和ResNet50上剪枝率为50%的情况下,精确度仅下降1.56%和0.11%;所提方法在精度损失一致下,ResNet50上具有更高剪枝率和更低计算量。     

选择性集成算法分类与比较

选择性集成是当前机器学习领域的研究热点之一。由于选择性集成属于NP"难"问题,人们多利用启发式方法将选择性集成转化为其他问题来求得近似最优解,因为各种算法的出发点和描述角度各不相同,现有的大量选择性集成算法显得繁杂而没有规律。为便于研究人员迅速了解和应用本领域的最新进展,本文根据选择过程中核心策略的特征将选择性集成算法分为四类,即迭代优化法、排名法、分簇法、模式挖掘法;然后利用UCI数据库的20个常用数据集,从预测性能、选择时间、结果集成分类器大小三个方面对这些典型算法进行了实验比较;最后总结了各类方法的优缺点,并展望了选择性集成的未来研究重点。     

An evaluation of new heuristics for the location of cross-docks distribution centers in supply chain network design

This paper addresses an evaluation of new heuristics solution procedures for the location of cross-docks and distribution centers in supply chain network design. The model is characterized by multiple product families, a central manufacturing plant site, multiple cross-docking and distribution center sites, and retail outlets which demand multiple units of several commodities. This paper describes two heuristics that generate globally feasible, near optimal distribution system design and utilization strategies utilizing the simulated annealing (SA) methodology. This study makes two important contributions. First, we continue the study of location planning for the cross-dock and distribution center supply chain network design problem. Second, we systematically evaluate the computational performance of this network design location model under more sophisticated heuristic control parameter settings to better understand interaction effects among the various factors comprising our experimental design, and present convergence results. The central idea of the paper is to evaluate the impact of geometric control mechanism vis-a-vis more sophisticated ones on solution time, quality, and convergence for two new heuristics. Our results suggest that integrating traditional simulated annealing with TABU search is recommended for this supply chain network design and location problem.     

A self-organizing neural network using ideas from the immune system to solve the traveling salesman problem

Most combinatorial optimization problems belong to the NP-complete or NP-hard classes, which means that they may require an infeasible processing time to be solved by an exhaustive search method. Thus, less expensive heuristics in respect to the processing time are commonly used. These heuristics can obtain satisfactory solutions in short running times, but there is no guarantee that the optimal solution will be found. Artificial Neural Networks (ANNs) have been widely studied to solve combinatorial problems, presenting encouraging results. This paper proposes some modifications on RABNET-TSP, an immune-inspired self-organizing neural network, for the solution of the Traveling Salesman Problem (TSP). The modified algorithm is compared with other neural methods from the literature and the results obtained suggest that the proposed method is competitive in relation to the other ones, outperforming them in many cases with regards to the quality (cost) of the solutions found, though demanding a greater time for convergence in many cases.     

An improved constraint satisfaction adaptive neural network for job-shop scheduling

This paper presents an improved constraint satisfaction adaptive neural network for job-shop scheduling problems. The neural network is constructed based on the constraint conditions of a job-shop scheduling problem. Its structure and neuron connections can change adaptively according to the real-time constraint satisfaction situations that arise during the solving process. Several heuristics are also integrated within the neural network to enhance its convergence, accelerate its convergence, and improve the quality of the solutions produced. An experimental study based on a set of benchmark job-shop scheduling problems shows that the improved constraint satisfaction adaptive neural network outperforms the original constraint satisfaction adaptive neural network in terms of computational time and the quality of schedules it produces. The neural network approach is also experimentally validated to outperform three classical heuristic algorithms that are widely used as the basis of many state-of-the-art scheduling systems. Hence, it may also be used to construct advanced job-shop scheduling systems.     

Applications of rule-base coverage measures to expert system evaluation

Often a rule-based system is tested by checking its performance on a number of test cases with known solutions, modifying the system until it gives the correct results for all or a sufficiently high proportion of the test cases. This method cannot guarantee that the rule-base has been adequately or completely covered during the testing process. We introduce an approach to testing of rule-based systems, which uses coverage measures to guide and evaluate the testing process. In addition, the coverage measures can be used to assist rule-base pruning and identification of class dependencies, and serve as the foundation for a set of test data selection heuristics. We also introduce a complexity metric for rule-bases.     

Design of efficient hybrid neural networks for flexible flow shop scheduling

Abstract: Although neural networks have been successfully used in performing pattern recognition, their application for solving optimization problems has been limited. In this paper we design a neural network to solve a well‐known combinatorial problem, namely the flexible flow shop problem. A key feature of our neural network design is the integration of problem structure and heuristic information in the network structure and solution. We compare the performance of our neural network with well‐known current heuristics with respect to solution quality. The results indicate that our approach outperforms the heuristics.     

Pruning recurrent neural networks for improved generalizationperformance

Determining the architecture of a neural network is an important issue for any learning task. For recurrent neural networks no general methods exist that permit the estimation of the number of layers of hidden neurons, the size of layers or the number of weights. We present a simple pruning heuristic that significantly improves the generalization performance of trained recurrent networks. We illustrate this heuristic by training a fully recurrent neural network on positive and negative strings of a regular grammar. We also show that rules extracted from networks trained with this pruning heuristic are more consistent with the rules to be learned. This performance improvement is obtained by pruning and retraining the networks. Simulations are shown for training and pruning a recurrent neural net on strings generated by two regular grammars, a randomly-generated 10-state grammar and an 8-state, triple-parity grammar. Further simulations indicate that this pruning method can have generalization performance superior to that obtained by training with weight decay.     

Classification based on association rules: A lattice-based approach

Classification plays an important role in decision support systems. A lot of methods for mining classification rules have been developed in recent years, such as C4.5 and ILA. These methods are, however, based on heuristics and greedy approaches to generate rule sets that are either too general or too overfitting for a given dataset. They thus often yield high error ratios. Recently, a new method for classification from data mining, called the Classification Based on Associations (CBA), has been proposed for mining class-association rules (CARs). This method has more advantages than the heuristic and greedy methods in that the former could easily remove noise, and the accuracy is thus higher. It can additionally generate a rule set that is more complete than C4.5 and ILA. One of the weaknesses of mining CARs is that it consumes more time than C4.5 and ILA because it has to check its generated rule with the set of the other rules. We thus propose an efficient pruning approach to build a classifier quickly. Firstly, we design a lattice structure and propose an algorithm for fast mining CARs using this lattice. Secondly, we develop some theorems and propose an algorithm for pruning redundant rules quickly based on these theorems. Experimental results also show that the proposed approach is more efficient than those used previously.     

Heuristics for Scheduling Data Requests Using Collective Communications in a Distributed Communication Network

The collective communication process studied here involves the ability of users to request information from diverse sources (both in terms of locality and data type) and receive this information before the users' deadlines expire. A model has been developed and examined for the case when the network is oversubscribed and not all requests in the system can be satisfied. This model is based upon systems that are in use or are planned for development. The heuristics presented for scheduling data traffic do not have explicit control of the network; the heuristics work with a network manager to configure and set up single-source, multiple-destination information channels that allow users to receive information. The goal is to create a near-optimal set of collective communication information channels that will satisfy the users' requests. Four components are employed to create value functions that are used to order the importance of the information channels for presentation to the network manager. In addition, two different weightings for the relative importance of different priority levels are considered. Simulation studies have been performed that evaluate the performance of the heuristics developed. It is shown that the heuristics perform well compared to the upper and lower bounds developed. These results can be used to develop techniques that can be deployed in actual military systems. Furthermore, the model and heuristics studied here can be applied to other domains, such as business applications that utilize the internet to obtain critical information.     

基于稀疏神经网络的图像超分辨率重建算法

部分基于深度学习的图像超分辨率重建算法通过扩展网络层的深度来提高网络模型的整体特征表达能力。然而,一味过度地扩展网络的深度会造成网络模型过参数化和复杂化,并且冗余的网络参数会增加特征表达的不稳定性。在LTH剪枝算法基础上改变权重参数并使用均衡学习策略,提出一种适用于图像超分辨率重建任务的神经网络非结构化剪枝算法RLTH。在不改变网络结构和不增加计算复杂度的前提下,通过搜索原始网络模型的最优稀疏子网络排除冗余参数带来的影响,在有限的参数资源中捕获更细粒度和丰富的图像特征,进而提高网络模型的整体特征表达能力。基于Set5、Set14和BSD100测试集的实验结果表明,与原始网络模型和应用LTH剪枝算法相比,应用RLTH算法获得的重建图像PSNR和SSIM均得到提升,且具有更丰富的细节特征,整体和局部轮廓更清晰。