The problem of bandwidth allocation in computer networks can be likened to the supply–demand problem in economics. This paper presents the economic generalized particle model(EGPM) approach to intelligent allocation of network bandwidth. EGPM is a significant extension and further development of the generalized particle model (GPM)[1]. The approach comprises two major components: (1) dynamic allocation of network bandwidth based on GPM; and (2) dynamic modulation of price and demands of network bandwidth. The resulting algorithm can be easily implemented in a distributed fashion. Pricing being the network control mechanism in EGPM is carried out by a tatonnement process. We dis-cuss the EGPM’s convergence and show that the approach is efficient in achieving the global Pareto optimum. Via simulations, we test the approach, analyze its parameters and compare it with GPM and a genetic-algorithm-based solution.

The Load Rebalancing Problem (LRP) that reassigns tasks to processors so as to minimize the maximum load arises in the context of dynamic load balancing. Many applications such as on Web based environment, parallel computing on clusters can be stated as LRP. Solving LRP successfully would allow us to utilize resources better and achieve better performance. However LRP has been proven to be NP-hard, thus generating the exact solutions in tractable amount of time becomes infeasible when the problems become large. We present a new nature-inspired approximation algorithm based on the Waterflow Particle Mechanics (W-PM) model to compute in parallel approximate efficient solutions for LRPs. Just like other Nature-inspired Algorithms (NAs) drawing from observations of physical processes that occur in nature, the W-PM algorithm is inspired by kinematics and dynamics of waterflow. The W-PM algorithm maps the classical LRP to the flow of water flows in channels by corresponding mathematical model in which all water flows flow according to certain defined rules until reaching a stable state. By anti-mapping the stable state, the solution to LRP can be obtained.

The development of the Internet of Things brings new opportunities and challenges for sensor networks. The scale of sensor networks tends to be larger. And the fusion rules need to be intelligent. In this paper, we propose a new Internet of Things group search optimizer (ITGSO) to solve intelligent information fusion problems in the high-dimensional multi-sensor networks. ITGSO is inspired by the latest research achievement about leader decision in Nature and works about social coordination, which mainly consists of three parts: basic group search optimizer, binary group search optimizer, and social decision model. With ITGSO, we need less time to obtain minimum Bayes cost than particle swarm optimization. And information of uncertain social intelligent problems can be fused. In this paper, we give the theoretical basic of ITGSO and proved its validity via mathematical analysis and simulation results.

Social Networks (SN) is an increasingly popular topic in artificial intelligence research. One of the key directions is to model and study the behaviors of social agents. In this paper, we propose a new computational model which can serve as a powerful tool for the analysis of SN. Specifically, we add to the traditional socio metric methods a novel analytical method in order to deal with social behaviors more effectively, and then present a new bio-inspired model, the coordination generalized molecule model (CGMM). The proposed analytical method for social behaviors and CGMM are combined to give an algorithm that can be used to solve complex problems in SN. Traditionally, SN models were mainly descriptive and were built at a very coarse level, typically with only a few global parameters, and turned out to be not sufficiently useful for analyzing social behaviors. In this work, we explore bio-inspired analytical models for analyzing social behaviors of intelligent agents. Our objective is to propose an effective and practical method to model intelligent systems and their behaviors in an open and complex unpredictable world.

Creative thinking, which plays an essential role in the progress of human society, has an outstanding problem-solving ability. This paper presents a novel creativity-oriented optimization model (COOM) and algorithm (COOA) inspired by the creative thinking process. At first, COOM is constructed by simplifying the procedure of creative thinking while retaining its main characteristics. And then, COOA is presented for continuous optimization problems. It is a realization of COOM. As a new nature-inspired algorithm, COOA is different from other similar algorithms in terms of the basic principle, mathematical formalization and properties. Features of the COOM and the corresponding algorithm include a powerful processing ability for the complex problems, namely high-dimensional, highly nonlinear and random problems. The proposed approach also has the advantages in terms of the higher intelligence, effectiveness, parallelism and lower computation complexity. The properties of COOA, including convergence and parallelism, are discussed in detail. The numerous simulations on the CEC-2013 real-parameter optimization benchmark functions’ problems have shown the effectiveness and parallelism of the proposed approach.

Nature has always been a muse for those who dream in art or science. As it goes, optimization algorithms inspired by nature have been widely used to solve various scientific and engineering problems because of their intelligence and simplicity. As a novel nature‐inspired algorithm, the crystal energy optimizer (CEO) is proposed in this article. The proposed CEO is motivated by the following general observation on lake freezing in nature: the dynamics of crystals have possession of parallelism, openness, local interactivity, and self‐organization. It stimulates us to extend a crystal dynamic model in physics to a generalized crystal energy optimizer for traveling salesman problems, so as to exploit the advantages of crystal dynamic system and to realize the aforementioned purposes. The proposed CEO has these advantages: (1) it has the ability to perform large‐scale distributed parallel optimization; (2) it can converge and avoid local optimum; and (3) it is flexible and easy to adapt to a wide range of optimization problems.

Due to the dynamic structure in network topology and absence of a centralized administration in management, a specific routing algorithm satisfying the demands of QoS is required indeed in mobile Ad Hoc networks. A novel Social Group Search Optimizer algorithm is pro-posed by improving the GSO algorithm to a dynamic and discrete algorithm through the introducing of social behaviors. SGSO is divided into search and prey parts, where “search” is on duty to find the optimal solution effectively and “prey” is responsible for adjusting the algorithm to the dynamic change of objective functions. Dynamic Coupling Level is used to divide the Ad Hoc network and corresponding approaches and models based on SGSO are applied to routing algorithm, including the decision factor and local routing table. The convergence and correctness of our algorithm are verified mathematically and extensive experiments have been conducted to evaluate the efficiency and effectiveness of the proposed mechanism in mobile Ad Hoc networks. The results show that SGSO improves packet delivery ratio and reduces average end-to-end latency effectively, especially for large-scale and high-dynamicnetworks.

為了緩解Internet網絡擁擠狀況,提高用戶訪問網站的響應速度,從技術上解決由于網絡帶寬小、用戶訪問量大、網點分布不均等原因所造成的用戶訪問網站響應速度慢的問題,提出了一種新的緩存資源分配方法——細胞優化算法。該算法是模仿自然細胞系統功能的一種智能優化方法,其通過模擬細胞內部結構和原理,對細胞核、細胞質的濃度、細胞間的親和度、細胞優化機制、細胞的動態演化過程建立數學模型。給出了算法的并行計算結構和步驟。最后,通過理論證明、仿真實驗與同類算法的比較,驗證了算法求解CDN緩,存資源分配問題的有效性。

隨著當今世界逐漸從信息化轉型為數據化，模式識別和數據挖掘等領域面臨越來越大的挑戰.爆炸式增大的數據量使得特征選擇過程成為大數據模式識別等領域必不可少的環節.受動物界資源爭奪行為啟發，在由特征選擇模型轉變為資源分配問題模型中加入個體的資源爭奪行為，提出多群體公平算法(multi-colony fairness algorithm, MCFA)對該行為進行評判和處理，用以取得更優的分配方案(即更優特征子集)，其有機融合隨機搜索和啟發式搜索，且將filter方法和wrapper方法相結合，降低計算量的同時獲得更高的分類準確率.對提出的多群體公平算法進行了分析，從理論上證明了算法的收斂性和有效性；UCI機器學習數據庫數據集與4種經典特征選擇算法:順序前向搜索算法(sequential forward selection, SFS)、順序后向搜索算法(sequential backward selection, SBS)、順序前向浮動搜索算法(sequential floating forward selection, SFFS)、順序后向浮動搜索算法(sequential floating backward selection, SBFS)和3種主流特征選擇算法：相關性-冗余度特征選擇算法(relevance-redundancy feature selection, RRFS)、最大相關最小冗余算法(minimal-redundancy-maximal-relevance, mRMR)、ReliefF算法的對比實驗表明，提出的多群體公平算法能夠有效選擇規模和性能都比較好的特征子集.

In order to solve complex practical problems, the model of deep learning can not be limited to models such as deep neural networks. To deepen the learning model, we must actively explore various depth models. Based on this, we propose a deep evolutionary algorithm, that is group competition cooperation optimization (GCCO) algorithm. Unlike the deep learning, in the GCCO algorithm, depth is mainly reflected in multi-step iterations, feature transformation, and models are complex enough. Firstly, the bio-group model is introduced to simulate the behavior that the animals hunt for the food. Secondly, according to the rules of mutual benefit and survival of the fittest in nature, the competition model and cooperation model are introduced. Furthermore, in the individual mobility strategy, the wanderers adopt stochastic movement strategy based on feature transformation to avoid local optimization. The followers adopt the variable step size region replication method to balance the convergence speed and optimization precision. Finally, the GCCO algorithm and the other three comparison algorithms are used to test the performance of the algorithm on ten optimization functions. At the same time, in the actual problem of setting up the Shanghai gas station the to improve the timely rate, GCCO algorithm achieves better performance than the other three algorithms. Moreover, Compared to the Global Search, the GCCO algorithm takes less time to achieve similar effects to the Global Search.