Artificial neural networks (ANNs) are commonly labelled as black-boxes, lacking interpretability. This hinders human understanding of ANNs' behaviors. A need exists to generate a meaningful sequential logic for the production of a specific output. Decision trees exhibit better interpretability and expressive power due to their representation language and the existence of efficient algorithms to generate rules. Growing a decision tree based on the available data could produce larger than necessary trees or trees that do not generalise well. In this paper, we introduce two novel multivariate decision tree (MDT) algorithms for rule extraction from an ANN: an Exact-Convertible Decision Tree (EC-DT) and an Extended C-Net algorithm to transform a neural network with Rectified Linear Unit activation functions into a representative tree which can be used to extract multivariate rules for reasoning. While the EC-DT translates the ANN in a layer-wise manner to represent exactly the decision boundaries implicitlylearned by the hidden layers of the network, the Extended C-Net inherits the decompositional approach from EC-DT and combines with a C5 tree learning algorithm to construct the decision rules. The results suggest that while EC-DT is superior in preserving the structure and the accuracy of ANN, Extended C-Net generates the most compact and highly effective trees from ANN. Both proposed MDT algorithms generate rules including combinations of multiple attributes for precise interpretation of decision-making processes.
翻译:人工神经网络(ANNS)通常被贴上黑盒子标签,缺乏解释性。这妨碍了人类对ANNS行为的理解。需要为生产特定产出产生有意义的连续逻辑。决策树因其代表语言和存在高效的逻辑来产生规则,因此显示更清楚和表达能力。在现有数据的基础上种植决策树可以产生大于必要的树木或树木,但不能全面概括。在本文中,我们引入了两种从ANN 提取规则的新颖的多变量决定树(MDT)算法:一种可撤销的决定树(EC-DT)和扩展的C-Net算法,将具有校正线单位激活功能的线性网络网络转换成具有代表性的树。这可以用来为推理提取多变规则。EC-DT将ANN值翻译成一个层次,以代表网络隐藏层所隐含的决策界限。 扩展 C-Net 网络继承了EC-DT的分立法方法,并与C5-Net的高级解释结构相结合,包括C-NDT的精准性解释,同时将EC-Q-Q-Q-Q-Qralalalalalation 和C-Lisalationalmas 建立高端的C-C-DDM-Lisals 和C-C-C-Lisalmaxxx 形成一个高级的高级的精准的精准的C-C-S-S-C-C-C-C-C-C-C-S-DDDDMLMLMLML)。