Abstract
We consider the problem of learning generalized first-order representations of concepts from a small number of examples. We augment an inductive logic programming learner with 2 novel contributions. First, we define a distance measure between candidate concept representations that improves the efficiency of search for target concept and generalization. Second, we leverage richer human inputs in the form of advice to improve the sample efficiency of learning. We prove that the proposed distance measure is semantically valid and use that to derive a PAC bound. Our experiments on diverse learning tasks demonstrate both the effectiveness and efficiency of our approach.
Highlights
We study the case of learning from few examples, of which one-shot learning is a special case (Lake et al, 2015)
We show the relation to Normalized Compression Distance (NCD) for plan induction settings
We aim to answer the following questions explicitly: (Q1) Is Guided One-shot Concept Induction (GOCI) effective in “one-shot” concept induction? (Q2) How sample efficient is GOCI compared to baselines? (Q3) What is the relative contribution of the novel scoring function versus human guidance toward performance?
Summary
We study the case of learning from few examples, of which one-shot learning is a special case (Lake et al, 2015). Concept learning has been considered as problem solving by reflection (Stroulia and Goel, 1994), mechanical compositional concepts (Wilson and Latombe, 1994), learning probabilistic programs (Lake et al, 2015), etc. The earliest form of knowledge injection can be found in explanation-based approaches (Shavlik and Towell, 1989), our work relates to preference-elicitation framework (Braziunas and Boutilier, 2006), which guides learning via human preferences as an inductive bias. In our problem setting, the interaction module that seeks human guidance to select the most useful constraints (detailed in section 3.2.3) is similar in spirit to interactive (knowledge guided) evolutionary mutation process
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.