An Approach to Combining Explanation-based and Neural Learning Algorithms

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Deep learning in neural networks: An overview (2014, methodology): “Compare earlier BP-based finetuning of NNs initialized by rules of propositional logic (Shavlik & Towell, 1989) (Section 5.6.1).”
End-to-end Differentiable Proving (2017, neutral): “Early proposals for neural-symbolic networks are limited to propositional rules ( e.g. , EBL-ANN [36], KBANN [37] and C-IL 2 P [38]).”
Lifelong Learning: A Case Study. (1995, methodology): “Neu ral networkbased methods [53, 18, 28, 65] basically initialize neural n etwork weights using domain knowledge, then train the network using conven tional neural network training algorithms.”
Combining Symbolic and Neural Learning (1994, background): “Various groups have found that knowledge-based neural networks train faster than do "standard" neural networks (Berenji, 1991; Oliver & Schneider, 1988; Omlin & Giles, 1992; Shavlik & Towell, 1989 ), presumably because the initial information is used to choose a good starting point for the network.”
Knowledge-Based Artificial Neural Networks (1994, result): “KBANN (Knowledge-Based Artificial Neural Networks) – the successor to our EBL-ANN algorithm [51] – is such a system.”
Embedding Logic Rules Into Recurrent Neural Networks (2019, background): “ENL-ANN [20], KBANN [21], and C-ILP [22], which are limited to propositional formulae, suggest that combing given rules with neural network structures lead to explanatory and knowledge acquisition.”