TRAINING INDUCTIVE LOGIC PROGRAMMING ENHANCED DEEP BELIEF NETWORK MODELS FOR DISCRETE OPTIMIZATION

摘要

System and method for training inductive logic programming enhanceddeep belief network models for discrete optimization are disclosed. The systeminitializes (i) a dataset comprising values and (ii) a pre-defined threshold,partitions the values into a first set and a second set based on the pre-definedthreshold. Using Inductive Logic Programming (ILP) engine and a domainknowledge associated with the dataset, a machine learning model is constructedon the first set and the second set to obtain Boolean features, and using theBoolean features that are being appended to the dataset, a deep belief network(DBN) model is trained to identify an optimal set of values between the firstsetand the second set. Using the trained DBN model, the optimal set of values aresampled to generate samples. The pre-defined threshold is adjusted based onthegenerated samples, and the steps are repeated to obtain optimal samples.