中文摘要：

提出一种改进的混合压缩相符预测器．混合压缩相符预测器采用两阶段压缩过程：先将部分序列样本压缩成模型以知识的形式保存；再将知识传递给后续样本用于置信预测．混合压缩相符预测器不仅能提高计算效率，还提供巧妙的邻近性度量，从而极大地提高了第二阶段的预测效率．以田纳西一伊斯曼化工过程为例，验证了该方法的有效性和高效性．

英文摘要：

Conformal predictor is extended to hybrid-compression conformal predictor （HCCP） in order to improve the computational efficiency. HCCP executes compression in two stages： a. a compression expert is assigned to compress part of the sequence of data; b. it transmits the extracted information to the successive transductive prediction. As a result, HCCP yields competitive computational efficiency, as well as maintaining the predictive efficiency, due to the ingenious proximity between the examples produced in the second stage. A case study of Tennessee Eastman Process was provided to illustrate the advantage of the proposed method.