Feng-Cheng Yang

WATER FLOW-LIKE ALGORITHM FOR OBJECT GROUPING PROBLEMS

ABSTRACT This paper presents a novice heuristic algorithm, Water Flow-like Algorithm (WFA), for solving discrete optimization problems, particularly the bin packing problems. WFA simulates solution agents as water flows traversing the terrain mapped from the objective function. Governed by the gravitation force, water flows from higher attitudes to lower ones. Driven by the fluid momentum, water flows adjust their compositions and directions against the landforms by splitting into and merging from other flows. Water flows are allowed to move upward to higher attitudes once they possess enough momentum to overcome the potential barrier. Mostly, at least one flow can travel to the lowest region of the terrain under the consideration. In the atmosphere, some water of a flow will evaporate and return to the ground by precipitation. Inspired by the water flowing of the nature, WFA is designed as an optimization algorithm performing the water flow splitting, merging, and dropping (precipitation) operations to traverse the solution space. The number of solution agents deployed is dynamically changing. WFA is an evolutionary algorithm involving four water flow operations: splitting and moving, merging, evaporation, and precipitation. The computational flow and the four operations are extensively discussed. In addition to general operations of WFA, specific operations for bin packing problems are presented. A designed problem and a benchmark problem from OR-Lib are used to test WFA and to compare results with other methods, such as GA, POS, and EM. Numerical results show that WFA outperforms others in solving these BPPs.

Numerical Analysis of the Kinematic Dexterity of Mechanisms

A general approach to numerical analysis of the kinematic dexterity of mechanisms is presented. Dextrous workspace problems are first defined and illustrated with examples. Composite workspaces are introduced to characterize both positioning and orienting capabilities of mechanisms. A numerical formulation and computer implementation that incorporates computer graphics and a numerical algorithm for solving systems of nonlinear equations are presented. Using the composite workspace formulation and the computer implementation, numerical techniques for dextrous workspace analysis are presented. Examples are given to illustrate the techniques developed.

Optimising resource portfolio planning for capital-intensive industries under process-technology progress

This paper addresses the problem of resource portfolio planning of firms in high-tech, capital-intensive manufacturing industries. In light of the strategic importance of resource portfolio planning in these industries, we offer an alternative approach to modelling capacity planning and allocation problems that improves the deficiencies of prior models in dealing with three salient features of these industries, i.e. fast technological obsolescence, volatile market demand, and high capital expenditure. This paper first discusses the characteristics of resource portfolio planning problems including capacity adjustment and allocation. Next, we propose a new mathematical programming formulation that simultaneously optimises capacity planning and task assignment. For solution efficiency, a constraint-satisfied genetic algorithm (CSGA) is developed to solve the proposed mathematical programming problem on a real-time basis. The proposed modelling scheme is employed in the context of a semiconductor testing facility. Experimental results show that our approach can solve the resource portfolio planning problem more efficiently than a conventional optimisation solver. The overall contribution is an analytical tool that can be employed by decision makers responding to the dynamic technological progress and new product introduction at the strategic resource planning level.

Superior/Inferior Segment-Discriminated Ant System for combinatorial optimization problems

The Ant Colony Optimization method is a heuristic algorithm for solving various optimization problems, particularly the combinatorial optimization problems. Traditional ant-optimization methods might encounter search stagnation owing to a biased pheromone map that is dominated by local optimal trails. To overcome this drawback and lower the number of solution constructions for finding the optima, this paper presents an improving ant-optimization system, the Superior/Inferior Segment-Discriminated Ant System (SDAS). This system proposes a segment-based pheromone update strategy to deposit pheromone on superior segments and withdraw pheromone from inferior ones. The method uses the control-chart technique to define superior and inferior limits to partition the constructed solutions into superior, inferior, and ordinary solutions. Inferior and superior segments are then extracted from the superior and inferior solutions by stochastic set operations. Since the pheromone map is not easily dominated by any local optimal trail, the solution search is more efficient and effective. Several benchmarks from the TSP-LIB and OR-LIB were used as sample problems to test the proposed system against other ant-optimization systems, including the AS, ACS, AS_rank, AS_elite, and MMAS. Numerical results indicated that the SDAS obtains solutions that are similar to or better than others. Maturity index for the pheromone map was discussed and experimental results showed that the proposed method was able to prolong the time for the map to maturity to avoid earlier search stagnation.

Water Flow-Like Optimization Algorithm for Multi-objective Continuous Optimization Problem

This paper presents a WFA for Multi-objective Continuous Optimization Problems. Namely WFA4MC. In order to prove WFA4MC performances precisely, this research proposes Correctness and Coverness to measure non-dominated solutions in ZDT functions. Besides, the Generational Distance is used in the comparison with other heuristic algorithms. The result showed that based on the same limit of the number of objective function calls, the WFA4MC outperform than others.

A time window rolling- and GA-based method for the ynamic dispatching problem in photolithography area

The operation requirements and constraints in the photolithography area considered include lot-priority, machine load balance for critical layers, machine bounded by critical layers, heterogeneous processing capability, reticle pilot run, and dummy wafer test. Under these considerations, this paper presents a time window rolling- and GA-based scheduling system to assign and schedule arrived and on-the-way wafer lots to the photolithography machines. Time window is defined when the scheduler is triggered by extending a time period to round in wafer lots on the way within this window. The scheduled jobs are then executed accordingly until the next scheduling event is triggered; and thus the widow rolls. The presented model proposed four minimization sub-goals to enhance the machine utilization and throughput while reducing cycle time and nonproductive time. To verify the proposed methods, a prototype system namely “Photolithography Area Simulation System,” was developed. In addition, four performance indices are proposed for evaluating the scheduling methods. Numerical tests were conducted using historical operation data from a Taiwanese DRAM manufacturing factory. Results showed that the proposed method outperformed the manual one and a simple FCFS scheduling method.

Intrusion Detection Based on Active Networks

The network security is getting more important due to the wide-spread computer viruses and increasing network attacks. Nowadays, more and more security mechanisms, such as firewalls and intrusion detection systems (IDS), are introduced to protect the network from malicious attacks. This paper proposes an agent and service based intrusion detection and response system for active network. In contrast to a traditional passive network, an active network gives the nodes programmable ability to exercise various active network technologies. The intrusion response, service deployment, and service update mechanisms are centered on this technology. The proposed model of intrusion detection and response system (IDRS) catches network attacks and responses to stop the attacks at the first time to reduce the damage. Detecting, reporting, and responding capabilities are all embedded and integrated in the proposed system. A prototype system is developed using a novel data mining technology (the support vector machine) to enhance the detection function. In addition, several experiments were conducted to verify the system and results showed that the system was able to effectively identify the intrusions and respond promptly. Experiments also showed that the support vector machine outperforms the competitive neural networks in identifying the intrusions.

A BOOLEAN ALGEBRA BASED RULE EXTRACTION ALGORITHM FOR NEURAL NETWORKS WITH BINARY OR BIPOLAR INPUTS

ABSTRACT Neural networks have been applied in various domain including science, commerce, medicine, and industry. However, The knowledge learned by a trained neural network is difficult to understand. This paper proposes a Boolean algebra based algorithm to extract comprehensible Boolean rules from supervised feed-forward neural networks to uncover the black-boxed knowledge. This algorithm is called the BAB-BB rule extraction algorithm, which stands for a Boolean algebra based rule extraction algorithm for neural networks with binary and bipolar inputs. Decomposition techniques and interval arithmetic are used in the algorithm. First, each neuron associated with its inputs is analyzed and a Boolean function, describing the activation rule from its inputs to the neuron, is derived. These Boolean functions are merged into an aggregated Boolean rule according to the network topology. The Boolean rule is then further simplified by Boolean algebra operations. During the rule extraction procedure, redundant hidden neurons can be detected and removed without affecting the original function of the neural network. Examples of unipolar and bipolar inputs are presented to demonstrate the use of our algorithm. Finally, the Exclusive OR problem is presented and solved by our algorithm. Results show that our BAB-BB algorithm is practicable and of high efficiency.

Resource allocation embedded line balancing problem and ant colony optimization method

This article introduces a resource allocation embedded assembly line balancing problem (RAELBP), whose solution construction involves operations of operator selection/assignment and task sequencing/grouping. The mathematical model of the RAELBP is rigorously defined. In addition, an ant colony optimization (ACO) method, resource allocation first ACO method, is developed for the problem. An artificial ant selects and assigns an operator to each workstation first and then sequentially assigns assembly tasks to each workstation subject to precedence constraints. Alternatives and options relating to the objective function and heuristic value evaluations are provided. A data generator is developed to generate benchmarks for the problem. A software application, RAELBP Solver, implementing the proposed methods is developed for numerical tests. Numerical tests are conducted for different computation modes on three benchmarks of various sizes. Results indicate that the proposed method is capable of solving the problem and achievements in obtaining non-dominated solutions are varied in different modes.

SUPPORT AND CONFIDENCE BASED RULE EXTRACTION METHOD FOR NEURAL NETWORKS

This paper presents a rule extraction method for competitive learning neural networks that are used for data clustering. First, a partition algorithm is used to divide attribute values into non-overlapped intervals. Consistency evaluation method adopted from rough set theory is used to partition attribute values. The generation of the set of adjoined intervals is controlled by the consistency evaluation against with the data distribution on the neural networks. By keeping the level of consistency, the set of adjoined intervals correctly reflects the data distribution on the networks. Second, instead of exhaustively traversing all combinations of the intervals to test possible rules, our method constructs the rules systematically and recursively from lower dimensions to higher ones. Using and adapting the techniques of evaluating amounts of support and confidence for an association rule, the constructed rules from our method are supported by the data clustering to the networks with adequate confidence. Finally, a rule reduction and merging algorithm is used to obtain a concise yet accurate set of rules. To verify the correctness of the constructed rules from our method, five benchmark problems are tested and results are compared. Comparison shows that the correctness of the rules generated from our method is more accurate than those from decision tree C4.5.