Crow search algorithm (CSA), as a new swarm intelligence algorithm that simulates the crows’ behaviors of hiding and tracking food in nature, performs well in solving many optimization problems. However, while handling complex and high-dimensional global optimization problems, CSA is apt to fall into evolutionary stagnation and has slow convergence speed, low accuracy, and weak robustness. This is mainly because it only utilizes a single search stage, where position updating relies on random following among individuals or arbitrary flight of individuals. To address these deficiencies, a CSA with multi-stage search integration (MSCSA) is presented. Chaos and multiple opposition-based learning techniques are first introduced to improve original population quality and ergodicity. The free foraging stage based on normal random distribution and Lévy flight is designed to conduct local search for enhancing the solution accuracy. And the following stage using mixed guiding individuals is presented to perform global search for expanding the search space through tracing each other among individuals. Finally, the large-scale migration stage based on the best individual and mixed guiding individuals concentrates on increasing the population diversity and helping the population jump out of local optima by moving the population to a promising area. All of these strategies form multi-level and multi-granularity balances between global exploration and local exploitation throughout the evolution. The proposed MSCSA is compared with a range of other algorithms, including original CSA, three outstanding variants of CSA, two classical meta-heuristics, and six state-of-the-art meta-heuristics covering different categories. The experiments are conducted based on the complex and high-dimensional benchmark functions CEC 2017 and CEC 2010, respectively. The experimental and statistical results demonstrate that MSCSA is competitive for tackling large-scale complicated problems, and is significantly superior to the competitors.

No abstract

No abstract

Cloud task scheduling and resource allocation (TSRA) constitute a core issue in cloud computing. Batch submission is a common user task deployment mode in cloud computing systems. In this mode, it has been a challenge for cloud systems to balance the quality of user service and the revenue of cloud service provider (CSP). To this end, with multi-objective optimization (MOO) of minimizing task latency and energy consumption, we propose a cloud TSRA framework based on deep learning (DL). The system solves the TSRA problems of multiple task queues and virtual machine (VM) clusters by uniting multiple deep neural networks (DNNs) as task scheduler of cloud system. The DNNs are divided into exploration part and exploitation part. At each scheduling time step, the model saves the best outputs of all scheduling policies from each DNN to the experienced sample memory pool (SMP), and periodically selects random training samples from SMP to train each DNN of exploitation part. We designed a united deep learning (UDL) algorithm based on this framework. Experimental results show that the UDL algorithm can effectively solve the MOO problem of TSRA for cloud tasks, and performs better than benchmark algorithms such as heterogeneous distributed deep learning (HDDL) in terms of task scheduling performance.