Search

The traditional matrix decomposition model cannot fully explored the intrinsic relationship between the user and the object in the mean, bias and characteristics. This paper proposed a fitting matrix model to improve the prediction performance by constructing the user and the item matrix to represent the characteristics of the user and the item respectively. The matrix decomposition model had the advantage of accuracy in the field of recommender system, but the gradient descent method, which was the most popular method to train parameters of model, had a slow convergence speed. To resolve the above defects, this paper considered to accelerate the convergence speed using the convergence of quasi Newton method, and named the proposed algorithm as fitting matrix and two orders fusion iterative (FAST) algorithm. The experimental results show that the FAST algorithm is better than the traditional non negative matrix decomposition (NMF), singular value matrix decomposition (SVD), and the regularized singular value matrix decomposition (RSVD). FAST algorithm has a decrease with regard to the mean absolute error (MAE) and the root mean square error (RMSE), and has a significant improvement in the iterative efficiency, which alleviates the problem that the accuracy is difficult to balance with the efficiency of the iteration. [ABSTRACT FROM AUTHOR]

Swarm intelligent algorithms are derived from the simulation of natural evolution or collective behavior of animals to seek solutions of complicated optimization problems by exploring and exploiting the search space efficiently and effectively. Through the analysis on the search characteristics of swarm intelligent algorithms, the concept of solution set diversity is introduced in this paper according to the infrastructural changes of solution sets in search processes. Two categories of fundamental search strategies, i.e. the diversification search and the intensification search, are then defined and their influences on the stagnation of solution sets evolution are investigated on the basis of the solution set diversity. It is proved in this paper that an intensification strategy inevitably leads candidates solution to a single solution, which is one of the main sources of stagnation; while a diversification strategy is able to reach any point of the coding space from each initial candidate solution, i.e., the whole searching space is the reachable region of the diversification strategy, but the convergence of the algorithm cannot be guaranteed. Three popular swarm intelligent algorithms, i.e. the canonical generic algorithm, the ant colony system and the discrete particle swarm optimization, are tested with a benchmark problem, and the results support the theoretical conclusions. [ABSTRACT FROM AUTHOR]

By using the differential evolution algorithm (DE) to solve multi-objective optimization problems, this paper proposed a Pareto optimal solution migration based differential evolution for multi-objective optimization (PSDEMO) to guarantee the diversity of Pareto optimal solution. It adopted the elitist strategy in the algorithm, and archived Pareto non-dominance solutions found in the evolution operation dynamically with the evolution process. In addition, it used all the non-dominance solutions in the archive to do migration operation after mutation and crossover operation of DE to increase the number and quality of non-dominated solutions. Compared with standard DE, simulation results show that the PSDEMO not only helps to improve the quantity of the Pareto non-dominance solution, but also has good balance keeping ability between the diversity and convergence. [ABSTRACT FROM AUTHOR]

Aiming at the premature convergence problem of particle swarm optimization algorithm, this paper proposes an improved particle swarm optimization algorithm with adaptive mutation. Based on the variance of the population' s fitness and the current optimal solution size, the algorithm determines the best current variation factor of particles. Motion direction of the particle is changed by using variation factor. Then, the particles move into the adjacent regions to search and determine the new individual extreme and global extreme value. Thus, local optimal solution is avoided. The simulation results show that the adaptive mutation particle swarm optimization algorithm enhances the ability to jump out of local optimal solution, and its global search ability has been improved. Its convergence speed is faster, and can effectively avoid the premature convergence problem. Although the running time is increased, it is still in the acceptable range. A little time of obtaining the global optimal solution spends is worth. [ABSTRACT FROM AUTHOR]

Focus on the electricity load forecasting in energy-intensive enterprise, a novel Boosting forecasting method based on geometric conversion is proposed. This method extends the AdaBoost algorithm to the regression problem. In this method, a regression problem such as load forecasting is converted to a binary classification problem through geometric operation. Then confidence-rated AdaBoost is implemented on this classification problem and an ensemble separating plane is obtained. This paper proves that such a plane is identical to a regression function for original load forecasting problem, and also proves that the proposed method has theoretical convergence. Simulation results state that compared to some popular single forecasting models, the proposed method improves the accuracy of the load forecasting in energy-intensive enterprise. [ABSTRACT FROM AUTHOR]