The availability of rich structured knowledge sources has spurred interest in exploring Knowledge Injection in Neural Networks (KINN) as a means of mitigating the challenges associated with current Deep Learning systems. This has led to the development of hybrid AI systems that combine the purely data-driven learning of the neural network models with an infusion of knowledge from external sources. Such KINN systems include the development of retrieval- augmented neural models, neuro-symbolic systems, and a plethora of combinations of NNs and knowledge graphs and structured knowledge bases. We invite papers for 1st workshop on Knowledge Injection in Neural Networks (KINN), co-located with CIKM 2021.
This workshop is centered around exploring efficient and scalable techniques for KINN, developing a robust set of benchmarks for evaluating KINN systems, and the challenges involved with KINN. Workshop topics include but not limited to the following:
(1) What are the desiderata of building such hybrid knowledge-aware neural networks?
(2) How/where should the knowledge injection in neural networks happen to build efficient and scalable hybrid models?
(3) How can knowledge injection mitigate huge data dependence, bias/trust issues, and improve the explainability of deep learning systems?
(4) How can knowledge injection help compensate the lack of implicit (or "tacit") human knowledge, i.e., the type of knowledge that can not be codified or conveyed verbally?
(5) How can the reconciliation of inherent NN knowledge acquired from data with the external structured knowledge happen?
(6) How to ensure credibility, robustness, and trustworthiness of external knowledge sources
(7) What kind of knowledge representations, formal languages, and neural architectures facilitate efficient hybrid models?
(8) What kind of deep learning applications would benefit from knowledge injection?
(9) What kind of Benchmarks and evaluation measures need to be developed for KINN?