论文部分内容阅读
长期以来大家认为人类认知尽管可以看成是非确定的推理计算过程,但它的知识表达、模型结构、及计算方法和概率统计理论在本质上是不同的,因此认知科学和概率统计方法存在巨大的鸿沟,过去两者基本上独立发展。近年来随着Bayesian概率统计模型研究的一系列突破性工作和认知过程本质的不断被发现和挖掘,两者的相关性和互补性逐渐突显出来。许多研究者认为认知是近似遵循概率统计推理原则的,一些研究工作显示两者的结合有可能对人工智能发展产生深远的影响。本文对当前统计认知理论及应用研究的现状进行系统的梳理,并结合自身的研究对它今后的发展提出自己的看法。
For a long time, people think that although human cognition can be regarded as a non-deterministic process of inference calculation, its knowledge expression, model structure, and calculation theory and probability theory are different in nature. Therefore, cognitive science and probability statistical methods exist Huge divide, the past two basically independent development. In recent years, as the series of groundbreaking work and the essence of cognitive process that Bayesian probabilistic statistical model studies are constantly being discovered and excavated, the correlation and complementarity of the two are gradually highlighted. Many researchers think that cognition follows the principles of probability and statistics reasonably. Some research work shows that the combination of the two may have a far-reaching impact on the development of artificial intelligence. This paper systematically sorts out the current status of statistical cognitive theory and applied research, and puts forward my own views on its future development with its own research.