"Stop replacing salt with sugar!'': Towards Intuitive Human-Agent Teaching

2025-09-29

AI Summary

Subjective tasks—such as recipe ingredient substitution—are challenging for few-shot learning due to severe data scarcity and the absence of objective ground truth. Method: This paper proposes an incremental human-in-the-loop teaching framework that integrates domain knowledge into few-shot learning. It synergizes symbolic knowledge reasoning with neural few-shot adaptation, introduces a semantic consistency–driven example selection strategy, and constructs Recipe1MSubs—a synthetic instructional dataset modeling intuitive human teaching behavior. Contribution/Results: The core innovation lies in incorporating interpretable, domain-specific constraints to guide model generalization, significantly improving adaptation efficiency under sparse supervision. Experiments demonstrate that the model achieves 50% of the performance attained by a fully supervised baseline trained on 50K samples—using only 100 high-quality, human-curated examples. This reduces both annotation effort and computational cost substantially, establishing a scalable, interpretable paradigm for low-resource subjective task learning.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Search and Optimization: Learning to Search
• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Humans quickly learn new concepts from a small number of examples. Replicating this capacity with Artificial Intelligence (AI) systems has proven to be challenging. When it comes to learning subjective tasks-where there is an evident scarcity of data-this capacity needs to be recreated. In this work, we propose an intuitive human-agent teaching architecture in which the human can teach an agent how to perform a task by providing demonstrations, i.e., examples. To have an intuitive interaction, we argue that the agent should be able to learn incrementally from a few single examples. To allow for this, our objective is to broaden the agent's task understanding using domain knowledge. Then, using a learning method to enable the agent to learn efficiently from a limited number of examples. Finally, to optimize how human can select the most representative and less redundant examples to provide the agent with. We apply our proposed method to the subjective task of ingredient substitution, where the agent needs to learn how to substitute ingredients in recipes based on human examples. We replicate human input using the Recipe1MSubs dataset. In our experiments, the agent achieves half its task performance after only 100 examples are provided, compared to the complete training set of 50k examples. We show that by providing examples in strategic order along with a learning method that leverages external symbolic knowledge, the agent can generalize more efficiently.

Problem

• Developing intuitive human-agent teaching for subjective tasks
• Enabling incremental learning from limited human demonstrations
• Optimizing example selection for efficient knowledge transfer

Innovation

• Human-agent teaching with demonstration examples
• Incremental learning from few examples
• Leveraging external knowledge for generalization