 |
ML Newsnew | past | best | rss | submit | about |
| Neural Datalog Through Time: Informed Temporal Modeling via Logic Specification (arxiv.org) | |
| 4 points by benrbray on Sep 2, 2020 | hide | past | pdf | 1 comment on HN | |
|
In plain words: Logical rules spell out how past events and facts combine, and those rules decide how a neural network is wired to track each fact's state over time. This rule-shaped network predicted future events better than a free-form neural model in synthetic and real-world tests. Abstract · Neural Datalog Through Time: Informed Temporal Modeling via Logical SpecificationLearning how to predict future events from patterns of past events is difficult when the set of possible event types is large. Training an unrestricted neural model might overfit to spurious patterns. To exploit domain-specific knowledge of how past events might affect an event's present probability, we propose using a temporal deductive database to track structured facts over time. Rules serve to prove facts from other facts and from past events. Each fact has a time-varying state---a vector computed by a neural net whose topology is determined by the fact's provenance, including its experience of past events. The possible event types at any time are given by special facts, whose probabilities are neurally modeled alongside their states. In both synthetic and real-world domains, we show that neural probabilistic models derived from concise Datalog programs improve prediction by encoding appropriate domain knowledge in their architecture. | |
| add comment on HN |
> Learning how to predict future events from patterns of past events is difficult when the set of possible event types is large. Training an unrestricted neural model might overfit to spurious patterns.
> To exploit domain-specific knowledge of how past events might affect an event's present probability, we propose using a temporal deductive database to track structured facts over time. Rules serve to prove facts from other facts and from past events. Each fact has a time-varying state---a vector computed by a neural net whose topology is determined by the fact's provenance, including its experience of past events. The possible event types at any time are given by special facts, whose probabilities are neurally modeled alongside their states.
> In both synthetic and real-world domains, we show that neural probabilistic models derived from concise Datalog programs improve prediction by encoding appropriate domain knowledge in their architecture.