Xionglin Luo

Learn More
Support vector machine is an effective classification and regression method that uses machine learning theory to maximize the predictive accuracy while avoiding overfitting of data. L2 regularization has been commonly used. If the training dataset contains many noise variables, L1 regularization SVM will provide a better performance. However, both L1 and L2(More)
The constraints of output variables, input variables and intermediate variables exist widely in chemical process control. The inconsistency in different constraints may make constrained model predictive controller have no feasible solutions, which will bring harmful effect to practical production. To ensure the implementation of model predictive control,(More)
Soft sensor software based on ANN (artificial neural network) using BP or RBF was developed to estimate unmeasured variables such as product quality online. Some important topics including how to determine the delay time, how to simulate the dynamic system were discussed and solved. We applied a 3 layers BP network to identify the delay time of nonlinear(More)
In chemical processes there commonly exist highly nonlinear processes such as polymerization, PH process. Traditional multiple model control only establishes linearized submodels on finite steady states, and the loss of linearized models on nonsteady states can't meet the requirement for rigorous models during transition. So this paper proposes a varying(More)
Domain adaptation problems arise when the data distribution in test domain is different from that in training domain. In this paper, we provide a new error bound for the domain adaptive regression problem. Inspired by the original ideas in 0-1 classification, firstly, an error bound with a large amount of samples of source domain can be got in a new scene(More)
Traditional supervised learning algorithms assume that the training data and the test data are drawn from the same probability distribution. But in many cases, this assumption is too simplified, and too harsh in light of modern applications of machine learning. So domain adaptation problems are proposed when the data distribution in test domain is different(More)
Traditional supervised learning algorithms assume that the training data and the test data are drawn from the same probability distribution. Considering the assumption is not hold in modern applications of machine learning, domain adaptation problems are proposed when the data distribution in test domain is different from that in training domain. This paper(More)