【摘要】 数据归纳推理是利用二元数码的前缀复杂性特征通过计算机递归操作进行推理的方法。它的哲学特征主要表现为:多重解释原理的数字化;奥卡姆剃刀原理的量化计算;算法信息的复杂性。由于数据归纳发现了二元数码字符串中存在着普遍的先验概率,因此它在解决休谟问题时将归纳逻辑推理转化为信息概率计算。只要概率计算过程中误差呈现收敛状态,理论上休谟问题就可以解决。但是,利用计算机进行二元数码的概率计算面临形式系统的不完备性,因而实践中休谟问题的解决还存在着计算机操作上的合理性问题。由于计算机操作上的合理性表现为算法优化问题,而算法优化又面临着"没有免费的午餐"定律的认识论制约,所以数据归纳推理还面临着语义的形式化表征、柯尔莫哥洛夫概率公理的合理性等一些问题。更多还原

【Abstract】 Data induction reasoning is a method of reasoning through computer recursive operations using the complexity of the prefix of the binary code. Its philosophical characteristics are mainly manifested in the digitization of multiple interpretation principles, the quantitative calculation of Occam’s razor principle and the complexity of algorithmic information. Since data induction found that there is a universal prior probability in binary code strings, it transforms inductive logic reasoning into information probability calculation when solving Hume’s problem. As long as the error converges in the process of probability calculation, Hume’s problem can be solved in theory. However, using computer to calculate the probability of binary number is faced with the incompleteness of formal system, so there is still the rationality of computer operation to solve Hume’s problem in practice. Because the rationality of computer operation is the problem of algorithm optimization, and algorithm optimization is restricted by the epistemology of “no free lunch” law, data inductive reasoning also faces some problems, such as the formal representation of semantics, the rationality of Kolmogorov probability axiom and so on.更多还原