A Neuro-Symbolic Framework for Sequence Classification with Relational and Temporal Knowledge
This work addresses a gap in neuro-symbolic AI for sequence classification with dynamic knowledge, but it is incremental as it builds on existing frameworks by adding temporal and relational dimensions.
The paper tackles the problem of knowledge-driven sequence classification by incorporating relational and temporal knowledge that changes over time, and it introduces a new benchmarking framework to evaluate multi-stage neuro-symbolic and neural-only architectures, highlighting challenges and shortcomings in this novel setting.
One of the goals of neuro-symbolic artificial intelligence is to exploit background knowledge to improve the performance of learning tasks. However, most of the existing frameworks focus on the simplified scenario where knowledge does not change over time and does not cover the temporal dimension. In this work we consider the much more challenging problem of knowledge-driven sequence classification where different portions of knowledge must be employed at different timesteps, and temporal relations are available. Our experimental evaluation compares multi-stage neuro-symbolic and neural-only architectures, and it is conducted on a newly-introduced benchmarking framework. Results demonstrate the challenging nature of this novel setting, and also highlight under-explored shortcomings of neuro-symbolic methods, representing a precious reference for future research.