University of Hertfordshire

By the same authors

AvE: Assistance via Empowerment

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Standard

AvE: Assistance via Empowerment. / Du, Yuqing; Tiomkin, Stas; Kıcıman, Emre; Polani, Daniel; Abbeel, Pieter; Dragan, Anca.

Advances in Neural Information Processing Systems 33 (NeurIPS 2020). ed. / H. Larochelle; M. Ranzato; R. Hadsell; M.F. Balcan; H. Lin. 2020.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Harvard

Du, Y, Tiomkin, S, Kıcıman, E, Polani, D, Abbeel, P & Dragan, A 2020, AvE: Assistance via Empowerment. in H Larochelle, M Ranzato, R Hadsell, MF Balcan & H Lin (eds), Advances in Neural Information Processing Systems 33 (NeurIPS 2020). NeurIPS 2020, Vancouver, Canada, 6/12/20.

APA

Du, Y., Tiomkin, S., Kıcıman, E., Polani, D., Abbeel, P., & Dragan, A. (2020). AvE: Assistance via Empowerment. In H. Larochelle, M. Ranzato, R. Hadsell, M. F. Balcan, & H. Lin (Eds.), Advances in Neural Information Processing Systems 33 (NeurIPS 2020)

Vancouver

Du Y, Tiomkin S, Kıcıman E, Polani D, Abbeel P, Dragan A. AvE: Assistance via Empowerment. In Larochelle H, Ranzato M, Hadsell R, Balcan MF, Lin H, editors, Advances in Neural Information Processing Systems 33 (NeurIPS 2020). 2020

Author

Du, Yuqing ; Tiomkin, Stas ; Kıcıman, Emre ; Polani, Daniel ; Abbeel, Pieter ; Dragan, Anca. / AvE: Assistance via Empowerment. Advances in Neural Information Processing Systems 33 (NeurIPS 2020). editor / H. Larochelle ; M. Ranzato ; R. Hadsell ; M.F. Balcan ; H. Lin. 2020.

Bibtex

@inproceedings{be5f689ab0d844d48c20db6e5f395509,
title = "AvE: Assistance via Empowerment",
abstract = "One difficulty in using artificial agents for human-assistive applications lies in the challenge of accurately assisting with a person{\textquoteright}s goal(s). Existing methods tend to rely on inferring the human{\textquoteright}s goal, which is challenging when there are many potential goals or when the set of candidate goals is difficult to identify. We propose a new paradigm for assistance by instead increasing the human{\textquoteright}s ability to control their environment, and formalize this approach by augmenting reinforcement learning with human empowerment. This task-agnostic objective preserves the person{\textquoteright}s autonomy and ability to achieve any eventual state. We test our approach against assistance based on goal inference, highlighting scenarios where our method overcomes failure modes stemming from goal ambiguity or misspecification. As existing methods for estimating empowerment in continuous domains are computationally hard, precluding its use in real time learned assistance, we also propose an efficient empowerment-inspired proxy metric. Using this, we are able to successfully demonstrate our method in a shared autonomy user study for a challenging simulated teleoperation task with human-in-the-loop training.",
author = "Yuqing Du and Stas Tiomkin and Emre Kıcıman and Daniel Polani and Pieter Abbeel and Anca Dragan",
year = "2020",
month = dec,
day = "12",
language = "English",
editor = "Larochelle, {H. } and Ranzato, {M. } and Hadsell, {R. } and { Balcan}, M.F. and Lin, {H. }",
booktitle = "Advances in Neural Information Processing Systems 33 (NeurIPS 2020)",
note = "NeurIPS 2020 : Thirty-fourth Conference on Neural Information Processing Systems ; Conference date: 06-12-2020 Through 12-12-2020",
url = "https://neurips.cc/Conferences/2020/",

}

RIS

TY - GEN

T1 - AvE: Assistance via Empowerment

AU - Du, Yuqing

AU - Tiomkin, Stas

AU - Kıcıman, Emre

AU - Polani, Daniel

AU - Abbeel, Pieter

AU - Dragan, Anca

N1 - Conference code: 34

PY - 2020/12/12

Y1 - 2020/12/12

N2 - One difficulty in using artificial agents for human-assistive applications lies in the challenge of accurately assisting with a person’s goal(s). Existing methods tend to rely on inferring the human’s goal, which is challenging when there are many potential goals or when the set of candidate goals is difficult to identify. We propose a new paradigm for assistance by instead increasing the human’s ability to control their environment, and formalize this approach by augmenting reinforcement learning with human empowerment. This task-agnostic objective preserves the person’s autonomy and ability to achieve any eventual state. We test our approach against assistance based on goal inference, highlighting scenarios where our method overcomes failure modes stemming from goal ambiguity or misspecification. As existing methods for estimating empowerment in continuous domains are computationally hard, precluding its use in real time learned assistance, we also propose an efficient empowerment-inspired proxy metric. Using this, we are able to successfully demonstrate our method in a shared autonomy user study for a challenging simulated teleoperation task with human-in-the-loop training.

AB - One difficulty in using artificial agents for human-assistive applications lies in the challenge of accurately assisting with a person’s goal(s). Existing methods tend to rely on inferring the human’s goal, which is challenging when there are many potential goals or when the set of candidate goals is difficult to identify. We propose a new paradigm for assistance by instead increasing the human’s ability to control their environment, and formalize this approach by augmenting reinforcement learning with human empowerment. This task-agnostic objective preserves the person’s autonomy and ability to achieve any eventual state. We test our approach against assistance based on goal inference, highlighting scenarios where our method overcomes failure modes stemming from goal ambiguity or misspecification. As existing methods for estimating empowerment in continuous domains are computationally hard, precluding its use in real time learned assistance, we also propose an efficient empowerment-inspired proxy metric. Using this, we are able to successfully demonstrate our method in a shared autonomy user study for a challenging simulated teleoperation task with human-in-the-loop training.

UR - https://papers.nips.cc/paper/2020/hash/30de9ece7cf3790c8c39ccff1a044209-Abstract.html

M3 - Conference contribution

BT - Advances in Neural Information Processing Systems 33 (NeurIPS 2020)

A2 - Larochelle, H.

A2 - Ranzato, M.

A2 - Hadsell, R.

A2 - Balcan, M.F.

A2 - Lin, H.

T2 - NeurIPS 2020

Y2 - 6 December 2020 through 12 December 2020

ER -