Agricultural warning classification model

作者：Gang Zhao; Guangli Liu^*

来源：Journal of Convergence Information Technology, 2011, 6(11): 54-59.

DOI：10.4156/jcit.vol6.issue11.7

• Agricultural safety
• Agricultural Warning
• Classification models
• Error bound
• Feature selection algorithm
• Kernel methods
• Leave-one-out
• Leave-one-out error
• Linear classifiers
• Newton'
• s methods
• Production problems
• Training data
• Uncertain labels
• Uncertainty problems

Agricultural safety is an important problem especially in China. A new gain warning method named leave-one-out classification is firstly introduced which can be derived directly by leave-one-out error bound. By kernel methods, linear classifier can be constructed. And agricultural warning feature selection algorithm is also proposed by minimizing leaver-one-out error bounds. In fact, agricultural warning problem has uncertain input or output. And by Newton';s method, algorithm can be efficiently performed. Common eave-one-out classification cannot solute uncertainty problem with probabilistic warning degrees or labels. Thus uncertainty leave-one-out classification model is proposed then which can deal with training data with uncertain labels. We interpreted the meaning of uncertainty and then derived the algorithm. The agricultural warning production problem is solved by our model and algorithm. Data experiment shows the new classification model plays a good performance.

• 出版日期2011