Along with the great success of deep neural networks, there is also growing concern about their black-box nature. The interpretability issue affects people's trust on learning systems. It related to many ethical problems, e.g., algorithmic discrimination. Moreover, a desired property for networks become powerful tools in other research fields, drug discovery and genomics. In this survey, we conduct comprehensive review network research. We first clarify definition as it has been used...

In this paper we investigate several self-adaptive mechanisms to improve our previous work on NSDE, which is a recent DE variant for numerical optimization. The methods originate from another variant, SaDE, but are remarkably modified and extended fit NSDE. And thus NSDE (SaNSDE) proposed NSDEpsilas performance. Three utilized in SaNSDE: self-adaptation two candidate mutation strategies, self-adaptations controlling scale factor F crossover rate CR, respectively. Experimental studies carried...

In this paper, we propose a multilevel cooperative coevolution (MLCC) framework for large scale optimization problems. The motivation is to improve our previous work on grouping based (EACC-G), which has hard-to-determine parameter, group size, in tackling problem decomposition. decomposer takes size as parameter divide the objective vector into low dimensional subcomponents with random strategy. MLCC, set of decomposers constructed strategy different sizes. evolution process divided number...

The first cooperative co-evolutionary algorithm (CCEA) was proposed by Potter and De Jong in 1994 since then many CCEAs have been successfully applied to solving various complex optimization problems. In applying CCEAs, the problem is decomposed into multiple subproblems, each subproblem solved with a separate subpopulation, evolved an individual evolutionary (EA). Through co-evolution of EA subpopulations, complete solution acquired assembling representative members from subpopulation....

Traditional multiobjective evolutionary algorithms face a great challenge when dealing with many objectives. This is due to high proportion of nondominated solutions in the population and low selection pressure toward Pareto front. In order tackle this issue, series indicator-based have been proposed guide search process However, single indicator might be biased lead converge subregion paper, multi-indicator-based algorithm for many-objective optimization problems. The algorithm, namely...

ABSTRACT This paper studies the dynamic interaction between net positions of traders and risk premiums in commodity futures markets. Short‐term position changes are driven mainly by liquidity demands noncommercial traders, while long‐term variation is primarily hedging commercial traders. These two components influence expected returns with opposite signs. The gains from providing commercials largely offset premium they pay for obtaining price insurance.

Recently, evolutionary multitasking (EMT) has been proposed in the field of computation as a new search paradigm, for solving multiple optimization tasks simultaneously. By sharing useful traits found along process across different tasks, performance on each task could be enhanced. The autoencoding-based EMT is recently algorithm. In contrast to most existing algorithms, which conduct knowledge transfer implicitly via crossover, it intends perform explicitly among form solutions, enables...

Constrained multiobjective optimization problems (CMOPs) involve both conflicting objective functions and various constraints. Due to the presence of constraints, CMOPs' Pareto-optimal solutions are very likely lying on constraint boundaries. The experience from constrained single-objective has shown that quickly obtain such an optimal solution, search should surround boundary feasible region infeasible sides. In this article, we extend idea cope with CMOPs and, accordingly, propose a novel...

ABSTRACT We use proprietary data on intraday transactions at a futures brokerage to analyze how implied leverage influences trading performance. Across all investors, is negatively related performance, due partly increased costs and forced liquidations resulting from margin calls. Defining skill out‐of‐sample, we find that relative performance differentials across unskilled skilled investors persist. Unskilled investors' amplifies losses lottery preferences the disposition effect. Leverage...

The capacitated arc routing problem (CARP) is a challenging combinatorial optimization with many real-world applications, e.g., salting route and fleet management. There have been attempts at solving CARP using heuristic meta-heuristic approaches, including evolutionary algorithms. However, almost all such formulate as single-objective although it usually has more than one objective, especially considering its applications. This paper studies multiobjective (MO-CARP). A new memetic algorithm...

<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> The capacitated arc routing problem (CARP) has attracted much attention during the last few years due to its wide applications in real life. Since CARP is NP-hard and exact methods are only applicable small instances, heuristic metaheuristic widely adopted when solving CARP. In this paper, we propose a memetic algorithm, namely algorithm with extended neighborhood search (MAENS), for MAENS...

Most reported studies on differential evolution (DE) are obtained using low-dimensional problems, e.g., smaller than 100, which relatively small for many real-world problems. In this paper we propose two new efficient DE variants, named DECC-I and DECC-II, high-dimensional optimization (up to 1000 dimensions). The algorithms based a cooperative coevolution framework incorporated with several novel strategies. strategies mainly focus problem decomposition subcomponents cooperation....

In this paper, we consider the scenario that a population-based algorithm is applied to numerical optimization problem and solution needs be presented within given time budget. Although wide range of algorithms, such as evolutionary particle swarm optimizers, differential evolution, have been developed studied under scenario, performance an may vary significantly from problem. This implies there inherent risk associated with selection algorithms. We propose that, instead choosing existing...

Class evolution, the phenomenon of class emergence and disappearance, is an important research topic for data stream mining. All previous studies implicitly regard evolution as a transient change, which not true many real-world problems. This paper concerns scenario where classes emerge or disappear gradually. A class-based ensemble approach, namely Class-Based Evolution (CBCE), proposed. By maintaining base learner each dynamically updating learners with new data, CBCE can rapidly adjust to...

Class imbalance learning tackles supervised problems where some classes have significantly more examples than others. Most of the existing research focused only on binary-class cases. In this paper, we study multiclass and propose a dynamic sampling method (DyS) for multilayer perceptrons (MLP). DyS, each epoch training process, every example is fed to current MLP then probability it being selected estimated. DyS dynamically selects informative data train MLP. order evaluate understand its...

Software defect prediction can help to allocate testing resources efficiently through ranking software modules according their defects. Existing models that are optimized predict explicitly the number of defects in a module might fail give an accurate order because it is very difficult exact due noisy data. This paper introduces learning-to-rank approach construct by directly optimizing performance. In this paper, we build on our previous work, and further study whether idea model...