The gravitational search algorithm (GSA) is a relatively novel swarm intelligence optimization technique which has been shown to be competitive to other population-based intelligence optimization algorithms. However, there is still an insufficiency that is the low convergence speed of the standard gravitational search algorithm, and its being stalled easily in the evolutionary process. Considering those problems, an improved gravitational search algorithm is presented. A strategy of chaotic opposition-based learning is employed to generate an initial population, which makes it possible for the algorithm to achieve a better initial population, thus accelerating the convergence speed. In addition, the method makes full use of the exploration ability of the search strategy of artificial bee colony algorithm to guide the algorithm to jump out of the likely local optima. The results of numerical simulation experiment on a suite of 13 benchmark functions demonstrate the effectiveness and superiority of the improved gravitational search algorithm.