Timezone: »
In machine teaching, a concept is represented by (and inferred from) a small number of labeled examples. Various teaching models in the literature cast the interaction between teacher and learner in a way to obtain a small complexity (in terms of the number of examples required for teaching a concept) while obeying certain constraints that are meant to prevent unfair collusion between teacher and learner. In recent years, one major research goal has been to show interesting relationships between teaching complexity and the VC-dimension (VCD). So far, the only interesting relationship known from batch teaching settings is an upper bound quadratic in the VCD, on a parameter called recursive teaching dimension. The only known upper bound on teaching complexity that is linear in VCD was obtained in a model of teaching with sequences rather than batches.This paper is the first to provide an upper bound of VCD on a batch teaching complexity parameter. This parameter, called STDmin, is introduced here as a model of teaching that intuitively incorporates a notion of ``importance'' of an example for a concept. In designing the STDmin teaching model, we argue that the standard notion of collusion-freeness from the literature may be inadequate for certain applications; we hence propose three desirable properties of teaching complexity and demonstrate that they are satisfied by STDmin.
Author Information
Farnam Mansouri (University of Toronto)
Hans Simon (Max-Planck Institute)

1978: Diploma (roughly equivalent to Masters) in Mathematics (Univ. of Saarbrücken) 1981: Doctor's degree in Computer Science (Univ. of Saarbrücken) 1989: "Habilitation degree" (Univ. of Saarbrücken) 1979-1989: Research Assistant at the Computer Science Department of the University of Saarbrücken 87/88: Temporary Professor Position at the Univ. of Darmstadt 89: Temporary Professor Position at the Univ. of Dortmund 89: Temporary Professor Position at the Univ. of Saarbrücken 1990-1997: Professor (tenured) at the Univ. of Dortmund 1997-2020: Professor (tenured) at the Ruhr-University-Bochum since August: 2020 Professor im Ruhestand (Emeritus) For more detailed information see: https://www.ruhr-uni-bochum.de/lmi/simon/work/cv.html
Adish Singla (MPI-SWS)
Sandra Zilles (zilles@cs.uregina.ca)
Related Events (a corresponding poster, oral, or spotlight)
-
2022 Poster: On Batch Teaching with Sample Complexity Bounded by VCD »
Dates n/a. Room
More from the Same Authors
-
2021 : Reward Poisoning in Reinforcement Learning: Attacks Against Unknown Learners in Unknown Environments »
Amin Rakhsha · Xuezhou Zhang · Jerry Zhu · Adish Singla -
2021 : Poster: Fair Clustering Using Antidote Data »
Anshuman Chhabra · Adish Singla · Prasant Mohapatra -
2021 : Reinforcement Learning Under Algorithmic Triage »
Eleni Straitouri · Adish Singla · Vahid Balazadeh Meresht · Manuel Rodriguez -
2021 : Reward Poisoning in Reinforcement Learning: Attacks Against Unknown Learners in Unknown Environments »
Amin Rakhsha · Xuezhou Zhang · Jerry Zhu · Adish Singla -
2022 Poster: Envy-free Policy Teaching to Multiple Agents »
Jiarui Gan · R Majumdar · Adish Singla · Goran Radanovic -
2022 Poster: Exploration-Guided Reward Shaping for Reinforcement Learning under Sparse Rewards »
Rati Devidze · Parameswaran Kamalaruban · Adish Singla -
2022 Poster: Provable Defense against Backdoor Policies in Reinforcement Learning »
Shubham Bharti · Xuezhou Zhang · Adish Singla · Jerry Zhu -
2021 : Deep Reinforcement Learning for Online Control of Stochastic Partial Differential Equations »
Erfan Pirmorad · Farnam Mansouri · Amir-massoud Farahmand -
2021 : Fair Clustering Using Antidote Data »
Anshuman Chhabra · Adish Singla · Prasant Mohapatra -
2021 : Fairness Degrading Adversarial Attacks Against Clustering Algorithms »
Anshuman Chhabra · Adish Singla · Prasant Mohapatra -
2021 Poster: Curriculum Design for Teaching via Demonstrations: Theory and Applications »
Gaurav Yengera · Rati Devidze · Parameswaran Kamalaruban · Adish Singla -
2021 Poster: Explicable Reward Design for Reinforcement Learning Agents »
Rati Devidze · Goran Radanovic · Parameswaran Kamalaruban · Adish Singla -
2021 Poster: On Blame Attribution for Accountable Multi-Agent Sequential Decision Making »
Stelios Triantafyllou · Adish Singla · Goran Radanovic -
2021 Poster: Teaching an Active Learner with Contrastive Examples »
Chaoqi Wang · Adish Singla · Yuxin Chen -
2021 Poster: Teaching via Best-Case Counterexamples in the Learning-with-Equivalence-Queries Paradigm »
Akash Kumar · Yuxin Chen · Adish Singla -
2020 Poster: Synthesizing Tasks for Block-based Programming »
Umair Ahmed · Maria Christakis · Aleksandr Efremov · Nigel Fernandez · Ahana Ghosh · Abhik Roychoudhury · Adish Singla -
2020 Poster: Task-agnostic Exploration in Reinforcement Learning »
Xuezhou Zhang · Yuzhe Ma · Adish Singla -
2019 Poster: Teaching Multiple Concepts to a Forgetful Learner »
Anette Hunziker · Yuxin Chen · Oisin Mac Aodha · Manuel Gomez Rodriguez · Andreas Krause · Pietro Perona · Yisong Yue · Adish Singla -
2019 Poster: Preference-Based Batch and Sequential Teaching: Towards a Unified View of Models »
Farnam Mansouri · Yuxin Chen · Ara Vartanian · Jerry Zhu · Adish Singla -
2019 Poster: Learner-aware Teaching: Inverse Reinforcement Learning with Preferences and Constraints »
Sebastian Tschiatschek · Ahana Ghosh · Luis Haug · Rati Devidze · Adish Singla -
2018 : Assisted Inverse Reinforcement Learning »
Adish Singla · Rati Devidze -
2018 Poster: Understanding the Role of Adaptivity in Machine Teaching: The Case of Version Space Learners »
Yuxin Chen · Adish Singla · Oisin Mac Aodha · Pietro Perona · Yisong Yue -
2018 Poster: Teaching Inverse Reinforcement Learners via Features and Demonstrations »
Luis Haug · Sebastian Tschiatschek · Adish Singla -
2018 Poster: Enhancing the Accuracy and Fairness of Human Decision Making »
Isabel Valera · Adish Singla · Manuel Gomez Rodriguez -
2017 Workshop: Teaching Machines, Robots, and Humans »
Maya Cakmak · Anna Rafferty · Adish Singla · Jerry Zhu · Sandra Zilles