会员体验
专利管家(专利管理)
工作空间(专利管理)
风险监控(情报监控)
数据分析(专利分析)
侵权分析(诉讼无效)
联系我们
交流群
官方交流:
QQ群: 891211   
微信请扫码    >>>
现在联系顾问~
热词
    • 2. 发明申请
    • ANALYSIS OF GENE EXPRESSION DATA FOR MULTI-CLASS PREDICTION
    • 用于多级预测的基因表达数据分析
    • WO2003094086A2
    • 2003-11-13
    • PCT/GB2003/001258
    • 2003-03-24
    • BIOTECH RESEARCH VENTURES PTE LIMITEDDENISON, ChristopherTAN, Patrick, Boon, OoiOOI, Chia, Huey
    • TAN, Patrick, Boon, OoiOOI, Chia, Huey
    • G06F19/00
    • G06K9/6231G06F19/20G06F19/24G06N3/126
    • The invention provides a method for identifying, from a pool of parameters whose values are known for a plurality of samples ("training samples") which are classified into multiple classes, a subset of parameters whose values in a sample of unknown class are likely to be predictive of the class of that sample, the method comprising: (a) providing a first generation population of individuals, each individual representing a candidate subset of parameters; (b) for each individual, evaluating its fitness to predict the class of a sample of unknown class by determining its ability to predict the class of a plurality of test samples of known class;(c) generating a second generation population of individuals from selected individuals in the first generation population, wherein said selected individuals are selected according to their fitness as evaluated in step (b); (d) repeating steps (b) and (c) to evaluate the fitness of the individuals of the second generation population and to generate and evaluate further generations until a termination condition is reached; and(e) selecting the fittest individual, as evaluated in step (b) or step (d), from the final generation or from among the fittest individuals of a plurality of generations, wherein the class prediction in step (b) uses a discriminant-based classifier. The invention also provides related uses, computer programs, apparatuses and systems. The invention is particularly applicable to the classification of biological samples on the basis of gene expression data. The classifier is preferably a maximum likelihood classifier.
    • 本发明提供了一种用于从被分类为多个类的多个样本(“训练样本”)已知值的参数池中识别未知类样本中的值很可能的参数子集的方法 预测该样本的类别,该方法包括:(a)提供第一代个体群体,每个个体代表参数的候选子集​​; (b)对于每个人,通过确定其预测已知类别的多个测试样本的类别的能力来评估其适合度来预测未知类别的样本的类别;(c)从所选择的生成第二代个体群体 第一代人群中的个体,其中所述选择的个体根据其在步骤(b)中评估的适合度来选择; (d)重复步骤(b)和(c)以评估第二代人群的个体的适应度,并产生和评估进一步的代数,直到达到终止条件; 以及(e)从步骤(b)或步骤(d)中评估的最适合的个体中选择多代人中最适合的个体之间的适合个体,其中步骤(b)中的类预测使用判别式 的分类器。 本发明还提供了相关用途,计算机程序,设备和系统。 本发明特别适用于基于基因表达数据的生物样品的分类。 分类器优选地是最大似然分类器。