A neural network construction method for surrogate modeling of physics-based analysis

Neural network;Machine learning;Network topology

Sung, Woong Je ; Aerospace Engineering

University:Georgia Institute of Technology

Department:Aerospace Engineering

关键词: Neural network;    Machine learning;    Network topology;   

Others  :  https://smartech.gatech.edu/bitstream/1853/43721/1/sung_woongje_201205_phd.pdf

美国|英语

来源: SMARTech Repository

【 摘 要 】

A connectivity adjusting learning algorithm, Optimal Brain Growth (OBG) was proposed. Contrast to the conventional training methods for the Artificial Neural Network (ANN) which focus on the weight-only optimization, the OBG method trains both weights and connectivity of a network in a single training process. The standard Back-Propagation (BP) algorithm was extended to exploit the error gradient information of the latent connection whose current weight has zero value. Based on this, the OBG algorithm makes a rational decision between a further adjustment of an existing connection weight and a creation of a new connection having zero weight. The training efficiency of a growing network is maintained by freezing stabilized connections in the further optimization process. A stabilized computational unit is also decomposed into two units and a particular set of decomposition rules guarantees a seamless local re-initialization of a training trajectory. The OBG method was tested for the multiple canonical, regression and classification problems and for a surrogate modeling of the pressure distribution on transonic airfoils. The OBG method showed an improved learning capability in computationally efficient manner compared to the conventional weight-only training using connectivity-fixed Multilayer Perceptrons (MLPs).