A Novel L-Fuzzy Concept Learning via Two-Way Concept-Cognitive Learning and Residuated Implication
Jinzhong Pang,Biao Zhang,Minghao Chen
DOI: https://doi.org/10.1007/s40815-024-01729-9
IF: 4.085
2024-05-18
International Journal of Fuzzy Systems
Abstract:Concept-cognitive learning (CCL) has been an active topic in artificial intelligence and cognitive computing. Two-way concept-cognitive learning (TCCL), an efficient cognitive mechanism to discover knowledge from an arbitrary information granule, has attracted attention to the problems of concept learning. Although TCCL has been widely adopted for cognitive and learning concepts, the existing studies still have some issues: learning concepts can only from the standard formal context, i.e., extent with discrete information, and ignoring the fuzziness of object information that is also important to precise cognitive and learning concepts. Hence, this paper proposes a novel L -fuzzy concept learning method via two-way concept-cognitive learning and fuzzy residual implication, namely, L -fuzzy concept-cognitive learning (LF-CCL). Specifically, we give the theoretical underpinnings of the L -fuzzy CCL method and a pair of dual cognitive operators are presented based on residual implication. Then L -fuzzy two-way granule spaces are constructed to explore the transformation of information granules and the generation of knowledge granules from arbitrary cognitive clues. Meanwhile, we analyze the CCL progress using L -fuzzy operators based on fixed points. Moreover, the corresponding algorithms are designed, and the decision value of this method in financial investment applications is explored through experimental analysis. Finally, numerical experiments verify the effectiveness and flexibility of the proposed method.
computer science, information systems,automation & control systems, artificial intelligence