Search

Angle of Attack (AOA) is a crucial parameter which directly affects the aerodynamic forces of an aircraft. The measurement of AOA is required to ensure a safe flight within its designed flight envelop. This paper intends to summarise a comprehensive survey on the measurement techniques and estimation methods for AOA, specifically in Unmanned Aerial Vehicle (UAV) applications. In the case of UAVs, weight constraint plays a major role as far as sensor suites are concerned. This results in selecting a suitable estimation method to extract AOA using the available data from the autopilot. The most feasible and widely employed AOA measurement technique is by using the Multi-Hole Probes (MHPs). The MHP measures the AOA regarding the pressure variations between the ports. Due to the importance of MHP in AOA measurement, the calibration methods for the MHP are also included in this paper. This paper discusses the AOA measurement using virtual AOA sensors, their importance and the operation. Keywords: Angle of attack estimation, Angle of attack measurement, Angle of attack vane, Five-hole probe, Multi-hole probe, Multi-hole probe calibration, Nine-hole probe, Seven-hole probe, Virtual angle of attack sensor

Recent research suggests that in vitro neural networks created from dissociated neurons may be used for computing and performing machine learning tasks. To develop a better artificial intelligent system, a hybrid bio-silicon computer is worth exploring. However, the performance of current hybrid bio-silicon design is still far from the silicon-based computer. One reason may be that the living neural network has many intrinsic properties, such as the random network connectivity, high network sparsity, and large neural and synaptic variability. These properties may lead to new design considerations and existing algorithms may need to be adjusted for living neural network implementation. This work investigates the impact of neural variation and random connections on the inference of learning algorithms. A two-layer hybrid bio-silicon platform is constructed and a three-stage design method is proposed for fast development of living neural network algorithms. Neural variation and dynamics are verified by fitting model parameters with biological experimental results. Random connections are generated under different connection probabilities to vary network sparsity. A multi-layer perceptron algorithm is tested with biological constraints on the MNIST dataset. The results show that a reasonable inference accuracy can be achieved when neural variations and random network connections are taken into account. A new adaptive pre-processing technique is proposed to ensure good learning accuracy with different living neural network sparsity.

This work proposes a new algorithm for training neural networks to solve the problems of feature selection and function approximation. The algorithm applies different weight constraint functions for the hidden and the output layers of a multilayer perceptron neural network. The LASSO operator is applied to the hidden layer; therefore, the training provides automatic selection of relevant features and the standard norm regularization function is applied to the output layer. Therefore, we propose a multi-objective training algorithm that is able to select the important features while solving the approximation problem.

Short-term traffic flow forecast plays an important role in intelligent transportation systems. Existing traffic flow prediction model used deep layer neural network, which can’t prevent over fitting, resulting in performance loss and lack of generalization ability. We propose a regularized LSTM model that fused recurrent dropout and max-norm weight constraint. We apply recurrent dropout to the recurrent connections of LSTM network, and use max-norm weight constraint to arrest the input weights not to grow very large. Simultaneously, we merge ADAM optimizer into our model. We use three datasets from different countries. In order to compare with the other researchers in the field of traffic flow prediction, we introduce same features and perform the same time interval prediction task. The experiment results show that our model has the lowest root mean square error and mean absolute error than the basic LSTM and other machine learning model including BP neural network, RNN, stacked autoencoder.

In the past, most researchers have paid attention to the storage space allocation problem in maritime container terminals, while few have studied this problem in rail&ndash, water intermodal container terminals. Therefore, this paper proposes a storage space allocation problem to look for a symmetry point between the efficiency and effectivity of rail&ndash, water intermodal container terminals and the unbalanced allocations and reallocation operations of inbound containers in the railway operation area, which are two interactive aspects. In this paper, a two-stage model on the storage space allocation problem is formulated, whose objective is to balance inbound container distribution and minimize overlapping amounts, considering both stacking principles, such as container departure time, weight and stacking height, and containers left in railway container yards from earlier planning periods. In Stage 1, a novel simulated annealing algorithm based on heuristics is introduced and a new heuristic algorithm based on a rolling horizon approach is developed in Stage 2. Computational experiments are implemented to verify that the model and algorithm we introduce can enhance the storage effect feasibly and effectively. Additionally, two comparison experiments are carried out: the results show that the approach in the paper performs better than the regular allocation approach and weight constraint is the most important influence on container storage.

Weight-constrained route planning (WRP) over static graphs has been extensively studied due to its wide application to transportation networks. However, real transportation networks often evolve over time and are thus modeled as time-dependent graphs. In this paper, we study the WRP problem over a large time-dependent graph by incorporating continuous time and weight functions into it. Most existing works regarding route planning over time-dependent graphs are based on the first-in-first-out (FIFO) property. Unfortunately, the FIFO property does not hold for our problem. To solve the problem, we propose two novel route planning algorithms, namely, a baseline algorithm and an advanced algorithm. Specifically, the advanced algorithm is even more efficient than the baseline algorithm, as the advanced algorithm incorporates a fast traversal scheme and tight bounds of time functions to terminate the traversal as early as possible. We confirm the effectiveness and efficiency of our algorithms by extensive experiments on real datasets.

Short texts have become a kind of prevalent source of information, and the classification of these short texts in various forms is valuable to many applications. However, most existing short text classification approaches circumvent the sparsity problem by extending short texts (or their feature representations) or exploiting additional information to adapt short texts to traditional text classification approaches. In this paper, we try to solve the sparsity problem of short text classification in a different direction: adapting the classifier to short texts. We propose a sparse representation short text classification method based on entropy weighted constraint. The main idea behind this study is to consider that the short texts are similar in potentially specific subspace. Specifically, we first introduce word embedding to represent the initial sparse representation dictionary, and then a fast feature subset selection algorithm is used to filter the dictionary. Again, we design an objective function based on sparse representation of entropy weight constraint. The optimal value of the objective function is obtained by Lagrange multiplier method. Finally, the distance between the short text to be classified and the short text in each class is calculated under the subspace, and the short text is classified according to three classification rules. Experiments over five datasets show that the proposed approach can effectively alleviate the problem of sparse feature of short text and is more efficient and effective than the existing short text classification method.

Stiffened conical shells in launch vehicles are very sensitive to various forms of imperfections. As a type of equivalent imperfections, several perturbation load approaches are used to investigate the influence of dimple-shape imperfections on the load-carrying capacity of stiffened conical shells. Firstly, the effect of axial location of dimple is examined by single perturbation load approach (SPLA), since the stiffness of stiffened conical shells varies along axial direction. Then, worst multiple perturbation load approach (WMPLA) is employed to find the lower bound of the collapse load of stiffened conical shells, and also provides the knowledge to determine the number of dimples in the multiple perturbation load approach (MPLA). After that, the optimization of stiffened conical shells for imperfection-insensitive design is carried out, where the equivalent MPLA is adopted during the optimization process to reduce the computational cost. Illustrative example indicates that stiffened conical shells exhibit more complicated imperfection sensitivity compared to cylindrical shells, and the proposed optimization framework can find an imperfection-insensitive design under structural weight constraint in an efficient manner.

In this paper, an efficient concurrent optimization method of macrostructures, and material microstructures and orientations is proposed for maximizing natural frequency. It is assumed that the macrostructure is composed of uniform material with the same microstructure but with various orientation. The bi-directional evolutionary structural optimization (BESO) method is applied to optimize the macrostructure and its material microstructure under a given weight constraint. Meanwhile, the optimality condition with respect to local material orientation is derived and embedded in the two-scale design of macrostructures and material microstructures. Numerical examples are presented to demonstrate the capability and effectiveness of the proposed optimization algorithm. The results show that the current design of macrostructures, material microstructures, and local material orientation greatly improves structural dynamic performance.

A common strategy to discovering cancer associated copy number aberrations (CNAs) from a cohort of cancer samples is to detect recurrent CNAs (RCNAs). Although the previous methods can successfully identify communal RCNAs shared by nearly all tumor samples, detecting subgroup-specific RCNAs and their related subgroup samples from cancer samples with heterogeneity is still invalid for these existing approaches. In this paper, we introduce a novel integrated method called HetRCNA, which can identify statistically significant subgroup-specific RCNAs and their related subgroup samples. Based on matrix decomposition framework with weight constraint, HetRCNA can successfully measure the subgroup samples by coefficients of left vectors with weight constraint and subgroup-specific RCNAs by coefficients of the right vectors and significance test. When we evaluate HetRCNA on simulated dataset, the results show that HetRCNA gives the best performances among the competing methods and is robust to the noise factors of the simulated data. When HetRCNA is applied on a real breast cancer dataset, our approach successfully identifies a bunch of RCNA regions and the result is highly correlated with the results of the other two investigated approaches. Notably, the genomic regions identified by HetRCNA harbor many breast cancer related genes reported by previous researches.

The goal of the Nurse Scheduling Problem (NSP) is to find an assignment of nurses to shifts according to specific requirements. Given its practical relevance, many researchers have developed different strategies for solving several variants of the problem. One of such variants was recently addressed by an approach based on Answer Set Programming (ASP), obtaining promising results. Nonetheless, the original ASP encoding presents some intrinsic weaknesses, which are identified and eventually circumvented in this paper. The new encoding is designed by taking into account both intrinsic properties of NSP and internal details of ASP solvers, such as cardinality and weight constraint propagators. The performance gain of clingo and wasp is empirically verified on instances from ASP literature. As an additional contribution, the performance of clingo and wasp is compared to other declarative frameworks, namely SAT and ILP; the best performance is obtained by clingo running the new ASP encoding.

Gak-Gyu KimㆍSeong-Woo Kim ㆍSeong-Woo KimDepartment of Operations Research, Korea National Defense UniversityAs a special topic of the vehicle routing problems (VRP), VRPSDP extends the vehicle routing problem as considering simultaneous pickup and delivery for goods. The past studies have mainly dealt with a only weight constraint of a loading capacity for heterogeneous products. However. this study suggests VRPSDP considering separate loading area according to characteristics of loading species. The objective is to design a set of minimum distance routes for the vehicle routing assignment with independent capacity for heterogeneous species. And then we present a another HVRPSDP model which is easy to utilizes in a unique circumstance that is a guarantee of executing a task simultaneously from the various areas under restricted time and raising an application of vehicles that returns at the depot for the next mission like the military group. The optimal results of the suggested mathematical models are solved by the ILOG CPLEX software ver. 12.4 that is provided by IBM company.

This paper addresses the constraints preventing full utilization of an aircraft’s cargo compartment. The fact that many sorties are flown with less than the maximum allowable weight or volume results in more airlift sorties being required to move the same amount of cargo. Various solutions are investigated to more efficiently fill the volume while simultaneously maximizing against the weight constraint. Solving the problem of the inefficient utilization of the airlift cargo compartment space can increase aircraft and aircrew availability, reduce fuel consumption for a given level of output, and reduce the number of flights required into constrained airfields. We recommend a pallet stacking solution moving cargo from Dover Air Force Base to Ramstein Air Force Base that we estimate would reduce C-17 sorties 31.4% when aggregated over a three calendar day period saving over eight million dollars per year.

Evaluating the allocation efficiency of agricultural flood and drought mitigation project is to investigate that whether the benefit brought by the project is effective from the economics angle of input-output. For the non-standardized weight distribution of the input-output indicators in the data envelopment analysis model, an improved data envelopment analysis model is proposed. Through constructing the decision-making unit and strengthening the weight constraint, the range of efficiency values is expanded and the weight of input-output indicators is standardized. In the end, this model is used to evaluate the allocation efficiency of agricultural flood and drought mitigation project of China's 31 provinces, municipalities and autonomous regions.

The Lee weight is more appropriate for some practical situations than the Hamming weight as it takes into account of the magnitude of each digit of the word. In this paper, we obtain a sufficient condition over the number of parity check digits for codes correcting simultaneously random and burst errors with Lee weight consideration. This sufficient condition is an extension of the Varshamov-Gilbert-Sacks bound for codes correcting simultaneously random and burst errors with Lee weight constraint.

Optimal problems are investigated in the present work in order to control the natural frequencies of a torsional shaft system including the total weight constraint and effects of tuned mass dampers. Maier objective functional is used. Pontryagin’s Maximum Principle is employed to derive necessary optimality conditions of the optimal problems. Numerical simulations are performed to study effects of tuned mass dampers on controlling natural frequencies as well as minimizing the system’s weight. Advantages of the proposed method are also discussed.

This paper deals with a variant of minimum spanning tree problem with multiple constraints, which includes degree, weight and budgeting constraints simultaneously together. To model the problem, a zero-one programming is incorporated. An exact solution procedure called pattern recognition technique-based lexi-search algorithm is developed. A suitable numerical illustration is given to check the applicability of the developed algorithm. Furthermore, the algorithm is programmed in C and tested with randomly generated hard instances, computational results are also reported. The overall results reveal that the proposed algorithm is fairly proficient in the sense of both acquiring the optimal solutions and computational time.

Much work has been done to clarify the notion of metamodelling and new ideas, such as strict metamodelling, distinction between ontological and linguistic instantiation, unified modelling elements and deep instantiation, have been introduced. However, many of these ideas have not yet been fully developed and integrated into modelling languages with (concrete) syntax, rigorous semantics and tool support. Consequently, applying these ideas in practice and reasoning about their meaning is difficult, if not impossible. In this paper, we strive to add semantic rigour and conceptual clarity to metamodelling through the introduction of Nivel, a novel metamodelling language capable of expressing models spanning an arbitrary number of levels. Nivel is based on a core set of conceptual modelling concepts: class, generalisation, instantiation, attribute, value and association. Nivel adheres to a form of strict metamodelling and supports deep instantiation of classes, associations and attributes. A formal semantics is given for Nivel by translation to weight constraint rule language (WCRL), which enables decidable, automated reasoning about Nivel. The modelling facilities of Nivel and the utility of the formalisation are demonstrated in a case study on feature modelling.

Frequent graph pattern mining is one of the most interesting areas in data mining, and many researchers have developed a variety of approaches by suggesting efficient, useful mining techniques by integration of fundamental graph mining with other advanced mining works. However, previous graph mining approaches have faced fatal problems that cannot consider important characteristics in the real world because they cannot process both (1) different element importance and (2) multiple minimum support thresholds suitable for each graph element. In other words, graph elements in the real world have not only frequency factors but also their own importance, in addition, various elements composing graphs may require different thresholds according to their characteristics. However, traditional ones do not consider such features. To overcome these issues, we propose a new frequent graph pattern mining method, which can deal with both different element importance and multiple minimum support thresholds. Through the devised algorithm, we can obtain more meaningful graph pattern results with higher importance. We also demonstrate that the proposed algorithm has more outstanding performance compared to previous state-of-the-art approaches in terms of graph pattern generation, runtime, and memory usage.

Since frequent graph pattern mining was proposed, various approaches have been suggested by devising efficient techniques or integrating graph mining with other mining areas. However, previous methods have limitations that cannot reflect the following important characteristics in the real world to their mining processes. First, elements in the real world have their own importance as well as frequency, but traditional graph mining methods do not consider such features. Second, various elements composing graph databases may need thresholds different from one another according to their characteristics. However, since traditional approaches mine graph patterns on the basis of only a single threshold, losses of important pattern information can be caused. Motivated by these problems, we propose a new graph mining algorithm that can consider both different importance and multiple thresholds for each element of graphs. We also demonstrate outstanding performance of the proposed algorithm by comparing ours with previous state-of-the-art approaches.

Weighted Frequent Itemset Mining (WFIM) has been proposed as an alternative to frequent itemset mining that considers not only the frequency of items but also their relative importance. However, an important limitation of WFIM is that it does not consider how recent the patterns are. To address this issue, we extend WFIM to consider the recency of patterns, and thus present the Recent Weighted Frequent Itemset Mining (RWFIM). A projection-based algorithm named RWFIM-P is designed to mine Recent Weighted Frequent Itemsets (RWFIs) based on a novel upper-bound downward closure property. Moreover, an improved algorithm named RWFIM-PE is also proposed, which introduces a new pruning strategy named Estimated Weight of 2-itemset Pruning (EW2P) to prune unpromising candidate of RWFIs early. An experimental evaluation against a state-of-the-art WFIM algorithm on the real-world and synthetic datasets show that the proposed algorithms are highly efficient.

In this paper, we study the problem of computing a minimum cost Steiner tree subject to weight constraint in a series–parallel graph where each edge has a nonnegative integer cost and a nonnegative integer weight. We present a strongly polynomial time approximation scheme for this NP-complete problem.

In this paper, we study the problem of computing a minimum cost Steiner tree subject to a weight constraint in a Halin graph where each edge has a nonnegative integer cost and a nonnegative integer weight. We prove the NP-hardness of this problem and present a fully polynomial time approximation scheme for this NP-hard problem.

This research is concentrates on a very interesting work, the bin packing using hybrid genetic approach. The optimal and feasible packing of goods for transportation and distribution to various locations by satisfying the practical constraints are the key points in this project work. As the number of boxes for packing can not be predicted in advance and the boxes may not be of same category always. It also involves many practical constraints that are why the optimal packing makes much importance to the industries. This work presents a combinational of heuristic Genetic Algorithm (HGA) for solving Three Dimensional (3D) Single container arbitrary sized rectangular prismatic bin packing optimization problem by considering most of the practical constraints facing in logistic industries. This goal was achieved in this research by optimizing the empty volume inside the container using genetic approach. Feasible packing pattern was achieved by satisfying various practical constraints like box orientation, stack priority, container stability, weight constraint, overlapping constraint, shipment placement constraint. 3D bin packing problem consists of 'n' number of boxes being to be packed in to a container of standard dimension in such a way to maximize the volume utilization and in-turn profit. Furthermore, Boxes to be packed may be of arbitrary sizes. The user input data are the number of bins, its size, shape, weight, and constraints if any along with standard container dimension. This user input were stored in the database and encoded to string (chromosomes) format which were normally acceptable by GA. GA operators were allowed to act over these encoded strings for finding the best solution.

Here we examine data from a two-year research on the use of sticks as probes by two groups of wild capuchin monkeys (Sapajus libidinosus) in Serra da Capivara National Park (PI), Brazil. The use of sticks as probes is not usually observed among wild tufted capuchin (Sapajus spp.) populations, having been reported as a customary behavior only in SCNP groups. Probe tools are used to access small prey (insects or lizards) in rock cracks or tree trunks, or honey from wasps’ nests, and also to poke toads and poisonous snakes. Probe use is, so far, the only known case in which wild capuchins modify objects used as tools: branches are trimmed off, and tips, thinned. Tool preparation episodes involved up to four modification steps. Contrary to the stone tools used to crack hard nuts, probe tools don’t present any weight constraint for use by females, but there is nevertheless a strong male bias (97%) in the occurrence of probe tool use. There are also no diet biases that could explain this difference. Although males hunt more often than females, the latter main prey items are lizards, which are also the main targets of probe tool use. One possibility is that females may have fewer social opportunities to learn about probe tools.

Although still relatively new, sound sampling tags are proving to be powerful tools for studying echolocation in the wild. Compared to remote camera or listening systems that receive only fleeting glimpses of individuals, a tag follows an animal for periods of hours as it navigates and forages in its natural environment. Sensors can be added to sample movements, physiological parameters, and the environment of the animal. Such multisensor sound tags have been used extensively on odontocetes enabling fine-scale studies of prey search, selection, and capture behaviors that are expanding our knowledge of species traditionally thought of as hard to study. Given the severe weight constraint on tags for bats, short-range telemetry devices are still the only option for on-animal sound sensing. These have been used to study signal adaptation in captive animals but telemetry range is limited for outdoors use.

This paper provides a first insight into cost sharing rules for the continuous knapsack problem. Assuming a set of divisible items with weights from which a knapsack with a certain weight constraint is to be filled, different such (classes of) rules are discussed. Those - based on individual approvals of the items - optimally fill the knapsack and share the cost of the knapsack among the individuals. Using various reasonable properties of continuous knapsack cost sharing rules, we provide three characterization results.

International audience; Purpose : The focus of this paper is to enhance our observation and knowledge of interurban road freight transport. It explores some possibilities for improving roadside freight surveys, usually used to gather origin-destination data. Methodology : To achieve this, new questions are added to a standard roadside survey form, and the enhanced survey form is tested through two surveys. The new questions relate to currently unobserved variables: the vehicle volume occupied by freight; the method of organisation (double crew, relays) used by the carriers; the existence of speci c logistical imperatives; and drivers' compulsory breaks. Results : The questions on volume constraint and carrier organisation prove to be the most informative. These two questions are thus two promising areas for improvement in roadside freight surveys. The questions about speci c logistical imperatives and driver breaks prove less fruitful. Conclusion : The paper concludes that it is possible, through minor modi cations in data collection protocols, to signi cantly enhance observation quality in road freight transport activity.

Weighted pattern mining have been studied the importance of items. So far, in weight constraint based pattern mining, the weight has been considered the item's price. The price considered as the weight has a limit. The weight characteristic of weighted pattern mining should be considered case-by-case situation. Thus, we motivate by considering the special and individual case-by-case situation to find the exact frequent patterns. We propose how to set weight into frequent patterns mining with a case-by-case condition, called CWFM (closed contingency weighted pattern miming). Moreover, we devise information tables by using statistical and empirical data as strategic decision. In addition, we calculate the contingency weight using outer variables and values which are from information tables. CWFM extracts more meaningful and appropriate patterns reflected case-by-case situation. The proposed new mining method finds closed contingency weighted frequent patterns having a significance which represents the case-by-case situation.

We consider the weight-constrained minimum spanning tree problem which has important applications in telecommunication networks design.We discuss and compare several formulations. In order to strengthen these formulations, new classes of valid inequalities are introduced. They adapt the well-known cover, extended cover and lifted cover inequalities. They incorporate information from the two subsets: the set of spanning trees and the knapsack set. We report computational experiments where the best performance of a standard optimization package was obtained when using a formulation based on the well-known Miller-Tucker-Zemlin variables combined with separation of cut-set inequalities.

This Final Report documents the SBIR Phase I efforts that established concept feasibility of the Multi-Variant/Capability Next Generation Troop Seat (M-V/C NGTS). The M-V/C NGTS comprised three variants. Each of these variants specifically addressed, at the lightest weight possible, the three established performance criterions. The recent implementation of crashworthy troop seats over the last decade has identified that weight is the critical issue in their employment. By providing three variants the implementer will be able to choose the best solution for their application and not be limited by the weight constraint of a single solution. These SBIR Phase I efforts developed and fabricated three new variants of the NGTS that were significantly lighter and cheaper than the existing state-of-the-art.

In this paper we investigate the problem of selecting a committee consisting of k members from a list of m candidates. Each candidate has a certain cost or weight . The choice of the k -committee has to satisfy some budget or weight constraint: the sum of the weights of all committee members must not exceed a given value W . While the former part of the problem is a typical question in Social Choice Theory, the latter stems from Operations Research. The purpose of this paper is to link these two research fields: we first characterize reasonable ways of ranking sets of objects, i.e., candidates, and then develop efficient algorithms for the actual computation of optimal committees.

In this paper we present a multilevel hypergraph partitioning algorithm which satisfies not only vertex weight constraint but also edge weight constraint. In our multilevel paradigm, two FM (Fiduccia and Mattheyses, 1988) variants are performed alternately in the refinement stage. The first FM variant aims to symmetrize edge weight of the two partitioning blocks and the other aims to minimize cut size under dual constraints on both edge and vertex weight. Inheritance of edge weights by vertices is considered in the coarsening stage. Experimental results on ISPD-98 show that our algorithm satisfies both 10% and 2% dual constraints, producing solutions, under 10% dual constraints, with much more balanced edge weight than hMetis, MLPart, and FM in terms of imbalance proportion by 837.2%, 1039.1% and 762.9%, and reasonable cut size increment by 29.0%, 28.4% and -42.3%, run time increment by 109.2%, 61.8% and 43.7%, respectively.

A mathematical model for the free radical batch solution polymerization of methyl methacrylate that takes depropagation into account was developed. This model was then used to derive optimal temperature and initiator concentration policies to reduce residual monomer concentration to desired levels, producing at the same time a polymer with the desired number average molecular weight. An objective function was formulated to take account of the cost of the initiator with respect to the cost of time of reaction. It was observed that when the cost of initiator increased, optimal initiator concentration decreased whereas optimal temperature increased. Finally temperature reached a limiting value above which polymer with desired number average molecular weight could not be produced. These results give insight into the factors that determine the policies that could be employed in optimizing the operation of a reactor.

The configuration problem in manufacture is more complicated than most other fields. Therefore, the design of modeling and reasoning module for product configuration manager in manufacture is very important and complex. It is just what is mainly discussed in this paper. A framework of PCM in manufacture is put forward according to the expert system theories, and a series of formalized knowledge representations and a data-driven graph control strategy for the PCM are provided. In the knowledge representations, the weight constraint is defined to distinguish the origins of the parameter value. It can be used to simplify the consistency check and effectivity check of the parameter. Furthermore, the search space of the configuration problem can be also easily compressed. The control strategy strengthens the real-time interaction between the user and the PCM so as to decompose the reasoning process. The reasoning process of the control strategy is described in detail by the configuration reasoning graph which is defined in this paper.

Multi-hole collimators are the most important piece of instrumentation in determining the trade-off between spatial resolution and noise of nuclear medicine images. Recently, we discovered that the collimator-to-detector gap influences very much the hole-pattern artifacts and we also showed that there exist specific values of the gap that minimize the artifacts: here we call this phenomenon "penumbral masking" (PM). In this paper we studied the matter in more detail and we took into account PM in the problem of designing high-energy collimators, basically, by substituting the conventional hole-array pattern constraint with a mathematical expression of the PM effect. With this approach we found that an optimal solution to the problem always exist while, without the PM concept, for lead an optimal solution exists below 300 KeV, a suboptimal solution can be found within the range 300-320 KeV and no solution at all exist beyond 320 KeV, unless one violates one or both the penetration and the hole array pattern constraints. Thus, one is able to design artifacts-free collimators even if a somewhat lower sensitivity has to be accepted: 20% at 511 KeV with lead. Even with our approach, the weight constraint is a major problem when high-energies are implied but this does not affect the capability of reducing the hole-pattern artifacts. However, even when the theoretically optimal solution may not be feasible, this approach enlightens very much the design of high-energy collimators.

An approach for modelling configurable software product families (CSPFs) and for automated configuring of product individuals using the models is presented. It is based on a similar approach for configuring physical products. The conceptual foundation and syntax of the Koalish modelling language used for this purpose are defined. The language extends Koala, a component model and architecture description language, with explicit variation modelling mechanisms. Koalish is further provided a formal semantics by defining a translation from it to Weight Constraint Rule Language (WCRL), a form of logic programs. This allows using an existing inference tool for WCRL, smodels, to implement the reasoning needed in the configurator. The configurator is able to construct all valid product individuals, with respect to a Koalish model of a CSPF, that satisfy a given set of requirements. The implemented functionality of the configurator has been tested using small-scale toy examples, for which it performs adequately.

Logic programs $P$ and $Q$ are strongly equivalent if, given any program $R$, programs $P \cup R$ and $Q \cup R$ are equivalent (that is, have the same answer sets). Strong equivalence is convenient for the study of equivalent transformations of logic programs: one can prove that a local change is correct without considering the whole program. Lifschitz, Pearce and Valverde showed that Heyting's logic of here-and-there can be used to characterize strong equivalence for logic programs with nested expressions (which subsume the better-known extended disjunctive programs). This note considers a simpler, more direct characterization of strong equivalence for such programs, and shows that it can also be applied without modification to the weight constraint programs of Niemela and Simons. Thus, this characterization of strong equivalence is convenient for the study of equivalent transformations of logic programs written in the input languages of answer set programming systems dlv and SMODELS. The note concludes with a brief discussion of results that can be used to automate reasoning about strong equivalence, including a novel encoding that reduces the problem of deciding the strong equivalence of a pair of weight constraint programs to that of deciding the inconsistency of a weight constraint program.

The robustness and learning speeds of a neural network using the backpropagation algorithm are explored. An XOR experiment was performed on neural networks with one and two hidden layers. Robustness of the net was studied through removal of nodes and/or branches in hidden layers. It is observed that simulations with final output weights constrained to lie below a specified value provided superior performance, even when they were structurally damaged. Hence, for a two-hidden layer net, the weight constraint on interconnecting links yields a robust and faster-learning network. >

In this paper, we study the problem of computing a minimum cost Steiner tree subject to weight constraint in a series-parallel graph where each edge has a nonnegative integer cost and a nonnegative integer weight.We present a fully polynomial time approximation scheme for this NP-complete problem.

The problem considered is to partition an edge-valued, node-weighted, finite, connected, simple graph (called a network) into disjoint subgraphs, each of which has a total node weight that does not exceed a weight constraint, and so that the total value of edges interconnecting the subgraphs is minimized. This paper presents a new approach which is effective in solving the problem, and fast enough to be practical in finding optimal partitions of large networks. It uses the concept of “divide and conquer” to partition the problem into several subproblems, which can be efficiently solved one after the other by using depth-first search and branch-and-bound principle. The operations required under the algorithms are additions, comparisons, and logical operations on binary vectors.

We examine a class of stochastic deep learning models with a tractable method to compute information-theoretic quantities. Our contributions are three-fold: (i) we show how entropies and mutual informations can be derived from heuristic statistical physics methods, under the assumption that weight matrices are independent and orthogonally-invariant. (ii) We extend particular cases in which this result is known to be rigorously exact by providing a proof for two-layers networks with Gaussian random weights, using the recently introduced adaptive interpolation method. (iii) We propose an experiment framework with generative models of synthetic datasets, on which we train deep neural networks with a weight constraint designed so that the assumption in (i) is verified during learning. We study the behavior of entropies and mutual informations throughout learning and conclude that, in the proposed setting, the relationship between compression and generalization remains elusive.