TEL . 400-800-5566

Explainable AI by combining statistical and logical methods

Explainable AI aims to solve the transparency problem of AI, such that AI systems address how black box decisions are made. It inspects and understands each step of reasoning procedure. The combination of statistical learning and logical methods is one potential direction of solving AI interpretability. Statistical learning analyses data distribution together with explainable machine learning models, which gives explanation for the outputs of AI black box. Meanwhile, logical rules and reasoning handle the interpretability of AI decisions. Through well-designed learning methods, statistical and logical methods learn the latent meanings of AI systems and models so that interpretability can be achieved.

【结合统计与逻辑方法的人工智能可解释性】可解释人工智能旨在解决人工智能的透明性,即为人工智能系统如何做黑盒决策,其主要审查并且试图理解人工智能做决策时的每个步骤。统计学习和规则逻辑方法的结合是可解释人工智能的一个潜在的方向。统计学习旨在分析数据分布以及结合可解释机器学习模型来针对人工智能黑盒和输出进行解释,同时,规则逻辑推理针对人工智能给出决策上的解释性。通过设计学习方法,用统计和规则逻辑方法学习人工智能潜在含义,从而达到可解释性。
展开更多
业务咨询

loading...