{"responseHeader":{"status":0,"QTime":7,"params":{"q":"{!q.op=AND}id:\"102825\"","hl":"true","hl.simple.post":"","hl.fragsize":"5000","fq":"!embargo_tdt:[NOW TO *]","hl.fl":"ocr_t","hl.method":"unified","wt":"json","hl.simple.pre":""}},"response":{"numFound":1,"start":0,"docs":[{"file_name_t":"Soller-Automated_Detection.pdf","thumb_s":"/65/e0/65e0c474d9da38593ec13c81bd9d572872a45fb9.jpg","oldid_t":"compsci 10938","setname_s":"ir_computersa","restricted_i":0,"format_t":"application/pdf","modified_tdt":"2016-05-26T00:00:00Z","file_s":"/c7/d0/c7d09e4d6799d8edca0171a2fa390c1c70cf1b25.pdf","title_t":"Page 68","ocr_t":"CHAPTER 10 ARTIFICIAL NEURAL NETWORK MODELS In the previous section, linear and semilinear models did not perform well for delirium prediction from laboratory data. Alternative models produce nonlinear classification boundaries. Artificial neural networks are one architecture capable of modelling nonlinear classification boundaries. 10.1 Introduction to Artificial Neural Networks Within this dissertation, an artificial neural network is defined as a mapping G from a vector, x, in the input domain, Rn, to a vector 0 in the output domain. Static networks, the only kind of artificial neural network addressed within this dissertation, map the inputs presented at a fixed point in time to outputs; there is no memory of previous system inputs, outputs, or intermediate calculations. For the delirium classification problem presented in this dissertation, 0 = G (X), X E Rn, 0 E [ -1, 1]. (10.1) The neural network classifies a given input, x, as a positive case if 0 equals 1 and otherwise if 0 equals -1. In the context of the delirium study, 0 = 1 represents a high likelihood of a patient experiencing a delirious episode within their hospitalization. In practice, some values fall in between -1 and 1, and a cutpoint must be used for the classification. This dissertation defines weights as variable model parameters. Learning, neural network training, is defined as the process by which the algorithm determines weights to optimize a criterion within the mapping [76]. Examples of criteria include generalization, smoothness of interpolation, and classification of","id":102825,"created_tdt":"2016-05-26T00:00:00Z","parent_i":102961,"_version_":1642982670128381952}]},"highlighting":{"102825":{"ocr_t":[]}}}