Pith. sign in

REVIEW 5 minor 58 references

Decentralized algorithms learn near-optimal human–assistant coordination under information asymmetry at rates Õ(T^{3/4}) and Õ(√T), and (1−1/e) is the best efficient approximation factor.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

2026-07-10 13:45 UTC pith:YAWQRY7F

load-bearing objection First poly-time algorithms with sublinear (1-1/e)-assistance regret for online CIRL/assistance games, via a clean submodular reduction and matching hardness.

arxiv 2607.08012 v1 pith:YAWQRY7F submitted 2026-07-09 cs.LG cs.AIcs.GT

Provably Optimal Learning Algorithms for Assistance Games

Nivasini Ananthakrishnan , Mark Bedaywi , Michael I. Jordan , Stuart Russell , Nika Haghtalab This is my paper
classification cs.LG cs.AIcs.GT
keywords assistance gamesonline learningassistance regretsubmodular maximizationtracking regrethuman-AI coordinationcooperative inverse reinforcement learning
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

When a human who sees private preferences must repeatedly interact with an AI assistant that only sees the human’s actions, the two agents share a reward but lack a common language or joint planner. The paper shows that this online assistance game can still be solved efficiently: decentralized learning rules for both parties achieve (1−1/e)-approximate assistance regret that vanishes as T^{3/4}, and with a one-time shared encoding of policies the rate improves to the information-theoretically optimal √T. The same approximation factor is tight; nothing better is possible for polynomial-time algorithms unless RP=NP. The construction works by reducing joint policy search to online submodular maximization over a partition matroid and then decomposing the resulting regret into a stable centralized component plus an adaptive assistant component. A sympathetic reader cares because this supplies the first formal guarantee that assistive AI can learn to act on a human’s behalf without pre-specified communication protocols or centralized control.

Core claim

There exist polynomial-time decentralized algorithms for the human and the assistant that achieve (1−1/e)-approximate assistance regret of order T^{3/4}; with a shared map from sequences of human actions to assistant policies the same approximation improves to order √T (optimal up to logs); and no efficient algorithm can beat the (1−1/e) factor unless RP=NP.

What carries the argument

Assistance regret together with its stable–adaptive decomposition: the gap to the best joint policy in hindsight is bounded by the external regret of a centralized submodular maximizer, the number of policy switches that maximizer makes, and the assistant’s tracking regret against that moving target.

Load-bearing premise

The sequence of private preferences is fixed before the game starts and cannot react to what the human or assistant actually does.

What would settle it

An efficient algorithm that achieves sublinear α-approximate assistance regret for any constant α > 1−1/e on the offline max-coverage reduction used in the hardness proof, or a matching lower-bound instance showing that the √T rate is impossible without a shared encoding.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

0 major / 5 minor

Summary. The paper studies an online variant of assistance games (cooperative inverse RL) in which a human who observes a latent preference sequence θ^(t) and an assistant who observes only the human’s actions interact for T rounds under a common reward. It defines α-assistance regret against the best joint policy pair in hindsight and gives the first poly(M_H, M_A, N, T)-time decentralized algorithms that achieve (1-1/e)-approximate assistance regret Õ(T^{3/4}) (Theorem 4.1). With a shared encoding from human-action sequences to assistant policies the rate improves to Õ(√T), optimal up to logs (Theorem 4.2). A matching hardness result shows that any better approximation factor is intractable unless RP=NP (Theorem 4.3). The technical core is a reduction of joint policy optimization to online maximization of weighted threshold potentials over a partition matroid (the assistance matroid), a stability–adaptivity regret decomposition (Lemma 4.4), and standard black-box tools (RAOCO, POMER, Fixed-Share).

Significance. If the claims hold, the paper closes a long-standing open problem: the first computationally efficient, provably near-optimal learning algorithms for repeated assistance games. The reduction to online submodular maximization under matroid constraints is clean and reusable; the stability–adaptivity decomposition cleanly separates the coordination problem from the computational one; and the hardness result pins down the (1-1/e) barrier. The √T rate with shared randomness is information-theoretically tight (Proposition E.6), and the necessity of an oblivious adversary is proved rather than assumed (Appendix B). Complete proofs appear in the appendices, the algorithms are modular (any tracking-regret assistant works for the T^{3/4} result), and the discussion honestly lists the remaining open questions (dependence on action-space sizes, necessity of the shared map, richer interaction models). These are genuine strengths for a theory paper in cooperative multi-agent learning.

minor comments (5)
  1. Section 7 (Limitations) notes that the optimal √T rate relies on a pre-shared encoding φ. A short remark on whether a weaker form of synchronization (e.g., a short public random seed plus a fixed encoding of Π_A) would suffice would help readers assess practicality.
  2. In the proof of Proposition 6.1 / Appendix D.6 the Lipschitz constant of the concave relaxation is taken w.r.t. ℓ_1 and then converted to ℓ_2; a one-line reminder that G ≤ √(M_A M_H) would make the substitution into Theorem E.4 completely self-contained.
  3. Notation for the human policy sometimes appears as π_H and sometimes as the best-response map π*_H(θ); a consistent convention in Section 5 would improve readability.
  4. Appendix B (adaptive nature) is important for justifying the model; a forward pointer from the main-text model section would help readers who skip the appendix.
  5. A few minor typos appear (e.g., “asssistant” in the policy-space paragraph of Section 5.1; “the the assistant” in Lemma 4.4). A light copy-edit pass would clean them up.

Circularity Check

0 steps flagged

No circularity: assistance regret, matroid reduction, and hardness are independently defined and reduced to external published algorithms and Feige's max-k-coverage barrier.

full rationale

This is a theoretical online-learning paper whose central claims (Theorems 4.1–4.3) are obtained by (i) defining assistance regret independently of any algorithm (Definition 3.1), (ii) representing joint policies as a partition matroid and the objective as a weighted threshold potential (Definitions 5.2–5.3, Lemmas D.1–D.3, Proposition 5.4), (iii) decomposing decentralized regret into external regret, switches, and tracking regret (Lemma 4.4), and (iv) plugging in external published black-boxes (RAOCO/Salem et al. 2024, POMER/Agarwal et al. 2024, Fixed-Share tracking). The (1−1/e) factor is the classical submodular-maximization barrier, re-proved hard for assistance via a standard reduction from max-k-coverage (Lemma C.1, Feige 1998), not manufactured by self-definition. There are no fitted parameters, no self-referential normalizations, no uniqueness theorems imported from the same authors, and no renaming of an empirical pattern. The oblivious-nature assumption is an explicit modeling restriction with a matching linear-regret lower bound (Appendix B), not a circular premise. Derivation chain is self-contained against external benchmarks; score 0 is the correct outcome.

Axiom & Free-Parameter Ledger

0 free parameters · 4 axioms · 3 invented entities

Pure theoretical contribution. No free parameters are fitted. The load-bearing modeling choices are the oblivious adversary, finite discrete action/state spaces, common bounded reward, and the standard mathematical toolkit of online convex optimization, submodular maximization, and tracking regret. The invented entities are definitional scaffolding for the reduction, not physical postulates.

axioms (4)
  • domain assumption Nature is oblivious: the sequence θ(1)…θ(T) is fixed before interaction and independent of agents’ actions.
    Stated in Section 3; Appendix B proves that adaptive nature forces linear regret, so all sublinear claims rest on this restriction.
  • domain assumption Action spaces AH, AA and preference space Θ are finite; rewards lie in [0,1].
    Standard finite online-learning setting used throughout; poly-time claims are with respect to these sizes.
  • standard math Standard results on online convex optimization (POMER), Fixed-Share tracking regret, and RAOCO for weighted threshold potentials hold.
    Invoked via black-box citations (Agarwal et al. 2024, Herbster & Warmuth 1998, Salem et al. 2024); the paper only transfers their guarantees.
  • ad hoc to paper The value-of-assistance function is a weighted threshold potential whose concave relaxation is 1-Lipschitz (Lemmas D.2–D.3).
    Proved in the appendix from the definition of Vθ; enables the RAOCO reduction.
invented entities (3)
  • Assistance regret (α-approximate) no independent evidence
    purpose: Performance measure comparing cumulative reward to α times the best joint policy in hindsight.
    Definition 3.1; the central quantity bounded by the algorithms.
  • Assistance matroid no independent evidence
    purpose: Partition-matroid representation of assistant policies that makes joint optimization tractable.
    Definition 5.2; ground set AH×AA with capacity-1 parts per human action.
  • Value-of-assistance function Vθ no independent evidence
    purpose: Objective that maps independent sets (assistant policies) to the reward of the best-response human policy under preference θ.
    Definition 5.3; shown to be a weighted threshold potential, enabling the submodular reduction.

pith-pipeline@v1.1.0-grok45 · 35106 in / 2805 out tokens · 30000 ms · 2026-07-10T13:45:08.190847+00:00 · methodology

0 comments
read the original abstract

This paper studies an online variant of the assistance games framework, where an informed agent and an uninformed agent repeatedly interact over $T$ timesteps to optimize a common reward function. While the informed agent (the human) observes a latent state of the world, the uninformed agent (the assistant) observes only the human's actions. We provide the first provably efficient learning algorithms for repeated assistance games. We introduce the notion of assistance regret: the gap between the cumulative utility of interactions and that of the optimal joint policies in hindsight, which map latent states to action pairs. We present decentralized algorithms for both the human and the assistant that achieve a $(1-1/e)$-approximate assistance regret rate of $\widetilde{O}(T^{3/4})$, with runtime polynomial in the size of the action and state spaces. These algorithms are general; in particular, they accommodate any no-regret algorithm for the assistant. We prove that achieving a regret approximation factor better than $(1-1/e)$ is computationally intractable. Furthermore, we demonstrate how these generic no-regret algorithms can be tailored to a pseudo-decentralized setting -- using a shared random string -- to achieve a rate of $\widetilde{O}(T^{1/2})$, optimal up to logarithmic factors.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

58 extracted references · 58 canonical work pages · 6 internal anchors

  1. [1]

    Improved differentially private and lazy online convex optimization: Lower regret without smoothness requirements

    Naman Agarwal, Satyen Kale, Karan Singh, and Abhradeep Guha Thakurta. Improved differentially private and lazy online convex optimization: Lower regret without smoothness requirements. In Proceedings of the 41st International Conference on Machine Learning, volume 235 of Proceedings of Machine Learning Research, pages 343--361. PMLR, 2024

    work page 2024

  2. [2]

    Online learning over a finite action set with limited switching

    Jason Altschuler and Kunal Talwar. Online learning over a finite action set with limited switching. In Conference On Learning Theory, pages 1569--1573. PMLR, 2018

    work page 2018

  3. [3]

    Human expertise in algorithmic prediction

    Rohan Alur, Manish Raghavan, and Devavrat Shah. Human expertise in algorithmic prediction. Advances in Neural Information Processing Systems, 37: 0 138088--138129, 2024

    work page 2024

  4. [4]

    Online learning for adversaries with memory: price of past mistakes

    Oren Anava, Elad Hazan, and Shie Mannor. Online learning for adversaries with memory: price of past mistakes. In Advances in Neural Information Processing Systems, pages 784--792, 2015

    work page 2015

  5. [5]

    The allocation of decision authority to human and artificial intelligence

    Susan C Athey, Kevin A Bryan, and Joshua S Gans. The allocation of decision authority to human and artificial intelligence. In AEA Papers and Proceedings, volume 110, pages 80--84. American Economic Association 2014 Broadway, Suite 305, Nashville, TN 37203, 2020

    work page 2014

  6. [6]

    Does the whole exceed its parts? the effect of ai explanations on complementary team performance

    Gagan Bansal, Tongshuang Wu, Joyce Zhou, Raymond Fok, Besmira Nushi, Ece Kamar, Marco Tulio Ribeiro, and Daniel Weld. Does the whole exceed its parts? the effect of ai explanations on complementary team performance. In Proceedings of the 2021 CHI conference on human factors in computing systems, pages 1--16, 2021

    work page 2021

  7. [7]

    Is learning in games good for the learners? Advances in Neural Information Processing Systems, 36, 2024

    William Brown, Jon Schneider, and Kiran Vodrahalli. Is learning in games good for the learners? Advances in Neural Information Processing Systems, 36, 2024

    work page 2024

  8. [8]

    Mirror descent meets fixed share (and feels no regret)

    Nicolo Cesa-Bianchi, Pierre Gaillard, G \'a bor Lugosi, and Gilles Stoltz. Mirror descent meets fixed share (and feels no regret). Advances in Neural Information Processing Systems, 25, 2012

    work page 2012

  9. [9]

    Emergent communication at scale

    Rahma Chaabouni, Florian Strub, Florent Altch \'e , Eugene Tarassov, Corentin Tallec, Elnaz Davoodi, Kory Wallace Mathewson, Olivier Tieleman, Angeliki Lazaridou, and Bilal Piot. Emergent communication at scale. In International conference on learning representations, 2022

    work page 2022

  10. [10]

    Dependent randomized rounding via exchange properties of combinatorial structures

    Chandra Chekuri, Jan Vondr \'a k, and Rico Zenklusen. Dependent randomized rounding via exchange properties of combinatorial structures. In 2010 IEEE 51st Annual Symposium on Foundations of Computer Science, pages 575--584. IEEE, 2010

    work page 2010

  11. [11]

    On the np-completeness of finding an optimal strategy in games with common payoffs

    Francis Chu and Joseph Halpern. On the np-completeness of finding an optimal strategy in games with common payoffs. International Journal of Game Theory, 30: 0 99--106, 2001

    work page 2001

  12. [12]

    Collaborative prediction: Tractable information aggregation via agreement

    Natalie Collina, Ira Globus-Harris, Surbhi Goel, Varun Gupta, Aaron Roth, and Mirah Shi. Collaborative prediction: Tractable information aggregation via agreement. In Proceedings of the 2026 Annual ACM-SIAM Symposium on Discrete Algorithms (SODA), pages 4712--4798. SIAM, 2026

    work page 2026

  13. [13]

    Computing the optimal strategy to commit to

    Vincent Conitzer and Tuomas Sandholm. Computing the optimal strategy to commit to. In Proceedings of the 7th ACM conference on Electronic commerce, pages 82--90, 2006

    work page 2006

  14. [14]

    Strongly adaptive online learning

    Amit Daniely, Alon Gonen, and Shai Shalev-Shwartz. Strongly adaptive online learning. In International Conference on Machine Learning, pages 1405--1411. PMLR, 2015

    work page 2015

  15. [15]

    Human-algorithm collaboration: Achieving complementarity and avoiding unfairness

    Kate Donahue, Alexandra Chouldechova, and Krishnaram Kenthapadi. Human-algorithm collaboration: Achieving complementarity and avoiding unfairness. In Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency, pages 1639--1656, 2022

    work page 2022

  16. [16]

    A threshold of ln n for approximating set cover

    Uriel Feige. A threshold of ln n for approximating set cover. Journal of the ACM (JACM), 45 0 (4): 0 634--652, 1998

    work page 1998

  17. [17]

    A decision-theoretic model of assistance

    Alan Fern, Sriraam Natarajan, Kshitij Judah, and Prasad Tadepalli. A decision-theoretic model of assistance. Journal of Artificial Intelligence Research, 50: 0 71--104, 2014

    work page 2014

  18. [18]

    Learning to Communicate to Solve Riddles with Deep Distributed Recurrent Q-Networks

    Jakob N Foerster, Yannis M Assael, Nando de Freitas, and Shimon Whiteson. Learning to communicate to solve riddles with deep distributed recurrent q-networks. arXiv preprint arXiv:1602.02672, 2016

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  19. [19]

    Interpretation of optimal signals

    Michael Franke. Interpretation of optimal signals. New perspectives on games and interaction, pages 297--310, 2009 a

    work page 2009

  20. [20]

    Signal to act: Game theory in pragmatics

    Michael Franke. Signal to act: Game theory in pragmatics. University of Amsterdam, 2009 b

    work page 2009

  21. [21]

    Nash and correlated equilibria: Some complexity considerations

    Itzhak Gilboa and Eitan Zemel. Nash and correlated equilibria: Some complexity considerations. Games and Economic Behavior, 1 0 (1): 0 80--93, 1989

    work page 1989

  22. [22]

    The principles and limits of algorithm-in-the-loop decision making

    Ben Green and Yiling Chen. The principles and limits of algorithm-in-the-loop decision making. Proceedings of the ACM on human-computer interaction, 3 0 (CSCW): 0 1--24, 2019

    work page 2019

  23. [23]

    Designing algorithmic delegates: The role of indistinguishability in human-ai handoff

    Sophie Greenwood, Karen Levy, Solon Barocas, Hoda Heidari, and Jon Kleinberg. Designing algorithmic delegates: The role of indistinguishability in human-ai handoff. In Proceedings of the 26th ACM Conference on Economics and Computation, pages 306--336, 2025

    work page 2025

  24. [24]

    Cooperative inverse reinforcement learning

    Dylan Hadfield-Menell, Stuart J Russell, Pieter Abbeel, and Anca Dragan. Cooperative inverse reinforcement learning. Advances in neural information processing systems, 29, 2016

    work page 2016

  25. [25]

    The off-switch game

    Dylan Hadfield-Menell, Anca D Dragan, Pieter Abbeel, and Stuart Russell. The off-switch game. In AAAI Workshops, 2017

    work page 2017

  26. [26]

    Emergence of language with multi-agent games: Learning to communicate with sequences of symbols

    Serhii Havrylov and Ivan Titov. Emergence of language with multi-agent games: Learning to communicate with sequences of symbols. Advances in neural information processing systems, 30, 2017

    work page 2017

  27. [27]

    Adaptive algorithms for online decision problems

    Elad Hazan and Comandur Seshadhri. Adaptive algorithms for online decision problems. In Electronic colloquium on computational complexity (ECCC), volume 14, 2007

    work page 2007

  28. [28]

    Tracking the best expert

    Mark Herbster and Manfred K Warmuth. Tracking the best expert. Machine learning, 32 0 (2): 0 151--178, 1998

    work page 1998

  29. [29]

    Team decision theory and information structures in optimal control problems--part i

    Yu-Chi Ho et al. Team decision theory and information structures in optimal control problems--part i. IEEE Transactions on Automatic control, 17 0 (1): 0 15--22, 1972

    work page 1972

  30. [30]

    Regularized Conventions: Equilibrium Computation as a Model of Pragmatic Reasoning

    Athul Paul Jacob, Gabriele Farina, and Jacob Andreas. Regularized conventions: Equilibrium computation as a model of pragmatic reasoning. arXiv preprint arXiv:2311.09712, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  31. [31]

    Game dynamics connects semantics and pragmatics

    Gerhard J \"a ger. Game dynamics connects semantics and pragmatics. In Game theory and linguistic meaning, pages 103--117. Brill, 2007

    work page 2007

  32. [32]

    Game theory in semantics and pragmatics

    Gerhard J \"a ger. Game theory in semantics and pragmatics. Semantics: An international handbook of natural language meaning, 3: 0 2487--2516, 2012

    work page 2012

  33. [33]

    Online submodular welfare maximization: Greedy is optimal

    Michael Kapralov, Ian Post, and Jan Vondr \'a k. Online submodular welfare maximization: Greedy is optimal. In Proceedings of the twenty-fourth annual ACM-SIAM symposium on Discrete algorithms, pages 1216--1225. SIAM, 2013

    work page 2013

  34. [34]

    Emergent communication under varying sizes and connectivities

    Jooyeon Kim and Alice Oh. Emergent communication under varying sizes and connectivities. Advances in Neural Information Processing Systems, 34: 0 17579--17591, 2021

    work page 2021

  35. [35]

    Natural language from artificial life

    Simon Kirby. Natural language from artificial life. Artificial life, 8 0 (2): 0 185--215, 2002

    work page 2002

  36. [36]

    Iterated learning and the evolution of language

    Simon Kirby, Tom Griffiths, and Kenny Smith. Iterated learning and the evolution of language. Current opinion in neurobiology, 28: 0 108--114, 2014

    work page 2014

  37. [37]

    Assistancezero: Scalably solving assistance games

    Cassidy Laidlaw, Eli Bronstein, Timothy Guo, Dylan Feng, Lukas Berglund, Justin Svegliato, Stuart Russell, and Anca Dragan. Assistancezero: Scalably solving assistance games. In Forty-second International Conference on Machine Learning, 2025. URL https://openreview.net/forum?id=b9hVMJi0t2

    work page 2025

  38. [38]

    Emergent Multi-Agent Communication in the Deep Learning Era

    Angeliki Lazaridou and Marco Baroni. Emergent multi-agent communication in the deep learning era. arXiv preprint arXiv:2006.02419, 2020

    work page internal anchor Pith review Pith/arXiv arXiv 2006

  39. [39]

    Multi-Agent Cooperation and the Emergence of (Natural) Language

    Angeliki Lazaridou, Alexander Peysakhovich, and Marco Baroni. Multi-agent cooperation and the emergence of (natural) language. arXiv preprint arXiv:1612.07182, 2016

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  40. [40]

    Learning and approximating the optimal strategy to commit to

    Joshua Letchford, Vincent Conitzer, and Kamesh Munagala. Learning and approximating the optimal strategy to commit to. In Algorithmic Game Theory: Second International Symposium, SAGT 2009, Paphos, Cyprus, October 18-20, 2009. Proceedings 2, pages 250--262. Springer, 2009

    work page 2009

  41. [41]

    Ease-of-teaching and language structure from emergent communication

    Fushan Li and Michael Bowling. Ease-of-teaching and language structure from emergent communication. Advances in neural information processing systems, 32, 2019

    work page 2019

  42. [42]

    The geometry of logconcave functions and sampling algorithms

    L \'a szl \'o Lov \'a sz and Santosh Vempala. The geometry of logconcave functions and sampling algorithms. Random Structures & Algorithms, 30 0 (3): 0 307--358, 2007

    work page 2007

  43. [43]

    Decentralized stochastic control

    Aditya Mahajan and Mehnaz Mannan. Decentralized stochastic control. Annals of Operations Research, 241 0 (1): 0 109--126, 2016

    work page 2016

  44. [44]

    An efficient, generalized bellman update for cooperative inverse reinforcement learning

    Dhruv Malik, Malayandi Palaniappan, Jaime Fisac, Dylan Hadfield-Menell, Stuart Russell, and Anca Dragan. An efficient, generalized bellman update for cooperative inverse reinforcement learning. In International Conference on Machine Learning, pages 3394--3402. PMLR, 2018

    work page 2018

  45. [45]

    On team decision problems with nonclassical information structures

    Andreas A Malikopoulos. On team decision problems with nonclassical information structures. IEEE Transactions on Automatic Control, 68 0 (7): 0 3915--3930, 2022

    work page 2022

  46. [46]

    Decentralized stochastic control with partial history sharing: A common information approach

    Ashutosh Nayyar, Aditya Mahajan, and Demosthenis Teneketzis. Decentralized stochastic control with partial history sharing: A common information approach. IEEE Transactions on Automatic Control, 58 0 (7): 0 1644--1658, 2013

    work page 2013

  47. [47]

    Learning optimal strategies to commit to

    Binghui Peng, Weiran Shen, Pingzhong Tang, and Song Zuo. Learning optimal strategies to commit to. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, pages 2149--2156, 2019

    work page 2019

  48. [48]

    Team decision problems

    Roy Radner. Team decision problems. The Annals of Mathematical Statistics, 33 0 (3): 0 857--881, 1962

    work page 1962

  49. [49]

    Enhance the compositionality of emergent language by iterated learning

    Yi Ren, Shangmin Guo, Serhii Havrylov, Shay Cohen, and Simon Kirby. Enhance the compositionality of emergent language by iterated learning. In 3rd NeurIPS Workshop on Emergent Communication (EmeCom@ NeurIPS 2019). URL https://papers. nips. cc/book/advances-in-neural-information-processing-systems-32-2019, 2019

    work page 2019

  50. [50]

    "LazImpa": Lazy and Impatient neural agents learn to communicate efficiently

    Mathieu Rita, Rahma Chaabouni, and Emmanuel Dupoux. " lazimpa": Lazy and impatient neural agents learn to communicate efficiently. arXiv preprint arXiv:2010.01878, 2020

    work page internal anchor Pith review Pith/arXiv arXiv 2010

  51. [51]

    Online submodular maximization via online convex optimization

    Tareq Si Salem, G \"o zde \"O zcan, Iasonas Nikolaou, Evimaria Terzi, and Stratis Ioannidis. Online submodular maximization via online convex optimization. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 15038--15046, 2024

    work page 2024

  52. [52]

    Benefits of assistance over reward learning, 2020

    Rohin Shah, Pedro Freire, Neel Alex, Rachel Freedman, Dmitrii Krasheninnikov, Lawrence Chan, Michael D Dennis, Pieter Abbeel, Anca Dragan, and Stuart Russell. Benefits of assistance over reward learning, 2020

    work page 2020

  53. [53]

    Online learning and online convex optimization

    Shai Shalev-Shwartz et al. Online learning and online convex optimization. Foundations and Trends in Machine Learning , 4 0 (2): 0 107--194, 2012

    work page 2012

  54. [54]

    Lazy oco: Online convex optimization on a switching budget

    Uri Sherman and Tomer Koren. Lazy oco: Online convex optimization on a switching budget. In Conference on Learning Theory, pages 3972--3988. PMLR, 2021

    work page 2021

  55. [55]

    Bayesian modeling of human--ai complementarity

    Mark Steyvers, Heliodoro Tejeda, Gavin Kerrigan, and Padhraic Smyth. Bayesian modeling of human--ai complementarity. Proceedings of the National Academy of Sciences, 119 0 (11): 0 e2111547119, 2022

    work page 2022

  56. [56]

    Nash equilibria for an evolutionary language game

    Peter E Trapa and Martin A Nowak. Nash equilibria for an evolutionary language game. Journal of mathematical biology, 41 0 (2): 0 172--188, 2000

    work page 2000

  57. [57]

    Learning to Complement Humans

    Bryan Wilder, Eric Horvitz, and Ece Kamar. Learning to complement humans. arXiv preprint arXiv:2005.00582, 2020

    work page internal anchor Pith review Pith/arXiv arXiv 2005

  58. [58]

    Who leads and who follows in strategic classification? Advances in Neural Information Processing Systems, 34: 0 15257--15269, 2021

    Tijana Zrnic, Eric Mazumdar, Shankar Sastry, and Michael Jordan. Who leads and who follows in strategic classification? Advances in Neural Information Processing Systems, 34: 0 15257--15269, 2021

    work page 2021