Adaptive Teaching of Temporal Logic Formulas to Preference-based Learners

Authors: Zhe Xu, Yuxin Chen, Ufuk Topcu5061-5068

AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate our algorithm extensively under different classes of learners (i.e., learners with different preferences over hypotheses) and interaction protocols (e.g., nonadaptive and adaptive). Our results demonstrate the effectiveness of the proposed algorithm in teaching temporal logic formulas; in particular, we show that there are significant gains of teaching efficacy when the teacher adapts to feedback of the learner, or adapts to a (non-myopic) oracle.
Researcher Affiliation Academia Zhe Xu1, Yuxin Chen2, Ufuk Topcu3 1Arizona State University 2University of Chicago 3University of Texas at Austin
Pseudocode Yes Algorithm 1: Teaching of p LTLf Formulas with Integer Programming (TLIP) and Algorithm 2: Compute Demonstration
Open Source Code No The paper does not include an unambiguous statement about releasing the source code for the described methodology, nor does it provide a direct link to a code repository.
Open Datasets No The paper defines the hypothesis set and uses a simulated space. It does not mention using or providing access to a publicly available dataset with a specific link, DOI, or formal citation.
Dataset Splits No The paper describes randomly selecting initial and target hypotheses from a generated set and evaluates over "teaching sessions," but it does not specify explicit training, validation, and test dataset splits with percentages or sample counts.
Hardware Specification Yes timeout (TO) is 300 minutes (on a Mac Book with 1.40-GHz Core i5 CPU and 16-GB RAM)
Software Dependencies No The paper mentions using "highly-optimized IP solvers (Gurobi 2019)" but does not specify a version number for Gurobi or any other software dependencies.
Experiment Setup Yes The hypothesis set of p LTLf formulas are listed in Table 1. The set S of states is {0, 1, . . . , 10}. We randomly select both the initial hypothesis and the target hypothesis from the hypothesis set. The results are averaged over 10 teaching sessions, with the standard deviations listed in the supplementary material.