一种基于蚁狮优化的极限学习机

AN EXTREME LEARNING MACHINE BASED ON ANT LION OPTIMIZATION

极限学习机(Extreme Learning Machine,ELM)是一种高效率的单隐层前馈神经网络,由于其训练速度快与泛化性能好,在各个领域中都有广泛的应用。但是极限学习机随机生成输入权值与隐含层偏置矩阵,随机性影响训练模型的泛化性能与稳定性,降低模型分类的精度。为了解决这一问题,借鉴蚁狮优化算法中利用蚁狮种群中的多个个体进行并行寻优的能力,改进优化极限学习机的输入权值与隐含层偏置矩阵,得到一个分类精度更高模型。以UCI标准数据库中数据进行分类实验分析验证,实验结果表明,在5类UCI数据集上基于蚁狮优化的极限学习机(ALO-ELM)相比于PSO-ELM和SaDE-ELM具有更高的分类精度。

Extreme learning machine (ELM) is a high-efficiency single hidden layer feed forward neural network.Due to its fast training speed and good generalization performance,it is widely used in various fields.However,the extreme learning machine randomly generates the input weight and the implicit layer bias matrix.The randomness affects the generalization performance and stability of the training model and reduces the accuracy of the model classification.In order to solve this problem,the ant lion optimization algorithm was used to optimize the ability of multiple individuals in the ant lion population for parallel optimization,and the input weight and implicit layer bias matrix of the optimal extreme learning machine were improved to obtain a higher classification accuracy model.The algorithm was classified and experimentally verified by the data in the UCI standard database.The experimental results show that the extreme learning machine based on ant lion optimization (ALO-ELM) has higher classification accuracy than PSO-ELM and SaDE-ELM on the five types of UCI datasets.