Neural networks based on high-dimensional random feature generation have become popular under the notions extreme learning machine (ELM) and reservoir computing (RC). We provide an in-depth analysis of such networks w...Neural networks based on high-dimensional random feature generation have become popular under the notions extreme learning machine (ELM) and reservoir computing (RC). We provide an in-depth analysis of such networks with respect to feature selection, model complexity, and regularization. Starting from an ELM, we show how recurrent connections increase the effective complexity leading to reservoir networks. On the contrary, intrinsic plasticity (IP), a biologically inspired, unsupervised learning rule, acts as a task-specific feature regularizer, which tunes the effective model complexity. Combing both mechanisms in the framework of static reservoir computing, we achieve an excellent balance of feature complexity and regularization, which provides an impressive robustness to other model selection parameters like network size, initialization ranges, or the regularization parameter of the output learning. We demonstrate the advantages on several synthetic data as well as on benchmark tasks from the UCI repository providing practical insights how to use high-dimensional random networks for data processing.展开更多
The calculation method of forming limit of sheet metals based on M-K instability theory is proposed,and the method is applicable to different yield criterions and hardening models.The forming limit dia-grams of AA5754...The calculation method of forming limit of sheet metals based on M-K instability theory is proposed,and the method is applicable to different yield criterions and hardening models.The forming limit dia-grams of AA5754-O,AA6111-T4 aluminum alloy sheet and DP steel sheet under combined loading paths are obtained based on mixed hardening model with YLD2000-2D yield criterion proposed by Barlat in 2003 and L-C nonlinear kinematic hardening model proposed by Lemaitre and Chaboche.The results show that the forming limit diagram made up of limit strain(FLD-strain) is evidently influenced by the loading path.The forming limit diagram made up of limit stress(FLD-stress) is also influenced by loading path and it is not an only curve,which differs from the conventional view.The degree of the influence of loading path on FLD-stress is related with pre-strain.The larger the pre-strain is,the greater the influence of loading path on FLD-stress will be.The change of FLD-stress is small only when pre-strain is small.In addition,the hardening behavior of the material will influence the path-dependence of FLD-stress:The larger the proportion of kinematic hardening in the whole hard-ening is,namely the more obvious Bauschinger effect of the material,the greater the influence of loading path on FLD-stress will be.展开更多
文摘Neural networks based on high-dimensional random feature generation have become popular under the notions extreme learning machine (ELM) and reservoir computing (RC). We provide an in-depth analysis of such networks with respect to feature selection, model complexity, and regularization. Starting from an ELM, we show how recurrent connections increase the effective complexity leading to reservoir networks. On the contrary, intrinsic plasticity (IP), a biologically inspired, unsupervised learning rule, acts as a task-specific feature regularizer, which tunes the effective model complexity. Combing both mechanisms in the framework of static reservoir computing, we achieve an excellent balance of feature complexity and regularization, which provides an impressive robustness to other model selection parameters like network size, initialization ranges, or the regularization parameter of the output learning. We demonstrate the advantages on several synthetic data as well as on benchmark tasks from the UCI repository providing practical insights how to use high-dimensional random networks for data processing.
基金Natural Science Foundation for Key Program( 50935007 ) Natural Science Foundation (50805121)+3 种基金 National Basic Research Program of China (2010CB731701)Foundation for Fundamental Research of Northwestern Polytechnical University in PR China (NPU-FFR-JC20100229)Research Fund of the State Key Laboratory of Solidification Processing of Northwestern Polytechnical University in PR China (27-TZ-2010 ) the 111 Project (B08040)
基金Supported by the National Natural Science Foundation of China (Grant No. 50475004)
文摘The calculation method of forming limit of sheet metals based on M-K instability theory is proposed,and the method is applicable to different yield criterions and hardening models.The forming limit dia-grams of AA5754-O,AA6111-T4 aluminum alloy sheet and DP steel sheet under combined loading paths are obtained based on mixed hardening model with YLD2000-2D yield criterion proposed by Barlat in 2003 and L-C nonlinear kinematic hardening model proposed by Lemaitre and Chaboche.The results show that the forming limit diagram made up of limit strain(FLD-strain) is evidently influenced by the loading path.The forming limit diagram made up of limit stress(FLD-stress) is also influenced by loading path and it is not an only curve,which differs from the conventional view.The degree of the influence of loading path on FLD-stress is related with pre-strain.The larger the pre-strain is,the greater the influence of loading path on FLD-stress will be.The change of FLD-stress is small only when pre-strain is small.In addition,the hardening behavior of the material will influence the path-dependence of FLD-stress:The larger the proportion of kinematic hardening in the whole hard-ening is,namely the more obvious Bauschinger effect of the material,the greater the influence of loading path on FLD-stress will be.