Polynomial-Time Linear-Swap Regret Minimization in Imperfect-Information Sequential Games

No-regret learners seek to minimize the difference between the loss they cumulated through the actions they played, and the loss they would have cumulated in hindsight had they consistently modified their behavior according to some strategy transformation function. The size of the set of transformations considered by the learner determines a natural notion of rationality. As the set of transformations each learner considers grows, the strategies played by the learners recover more complex game-theoretic equilibria, including correlated equilibria in normal-form games and extensive-form correlated equilibria in extensive-form games. At the extreme, a no-swap-regret agent is one that minimizes regret against the set of all functions from the set of strategies to itself. While it is known that the no-swap-regret condition can be attained efficiently in nonsequential (normal-form) games, understanding what is the strongest notion of rationality that can be attained efficiently in the worst case in sequential (extensive-form) games is a longstanding open problem. In this paper we provide a positive result, by showing that it is possible, in any sequential game, to retain polynomial-time (in the game tree size) iterations while achieving sublinear regret with respect to all linear transformations of the mixed strategy space, a notion called no-linear-swap regret. This notion of hindsight rationality is as strong as no-swap-regret in nonsequential games, and stronger than no-trigger-regret in sequential games -- thereby proving the existence of a subset of extensive-form correlated equilibria robust to linear deviations, which we call linear-deviation correlated equilibria, that can be approached efficiently.

Paper

References (37)

Scroll for more · 25 remaining

Similar papers

Peer review

Reviewer Dcee7/10 · confidence 4/52023-07-01

Summary

Authors introduce a new class of correlated equilibria called linear-deviation correlated equilibria, which can be approached efficiently if all players attain sublinear linear-swap regret. They show LCEs are distinct from correlated equilibria and extensive-form correlated equilibria and the hardness of maximizing social welfare in LCEs. Finally, they show the difference in no-linear-swap-regret dynamics and no-trigger-regret dynamics to support their theoretical results on a small game.

Strengths

- Introduction of new correlated equilibrium class. - Technical contribution of a polynomial characterization of the set of all linear transformations from a sequence-form strategy polytope to itself. - General algorithm for minimizing linear-swap regret. - Shows relation to existing equilibrium classes. - Very clearly written paper that I expect will be built upon in the future.

Weaknesses

- This is a theoretical paper and it is not clear what would be the motivation to approach LCEs in practice, compared to another correlation classes.

Questions

- I think it would be of great pedagogical value to make a visual depiction of the different classes of equilibria similar to https://www.cs.cmu.edu/~ggordon/CE/ -- but I understand this might not be possible because the high dimensionality of the joint strategy space. - Do you think there exists a linear program to find LCEs? If not, why?

Rating

7: Accept: Technically solid paper, with high impact on at least one sub-area, or moderate-to-high impact on more than one areas, with good-to-excellent evaluation, resources, reproducibility, and no unaddressed ethical considerations.

Confidence

4: You are confident in your assessment, but not absolutely certain. It is unlikely, but not impossible, that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work.

Soundness

4 excellent

Presentation

4 excellent

Contribution

3 good

Limitations

Limitations are adequately addressed.

Reviewer qFDY6/10 · confidence 4/52023-07-02

Summary

This paper studies the convergence of uncoupled strategies to a weakened notion of equilibria called "linear-deviation correlated equilibrium". This equilibrium is reached when all players minimise the no-linear-swap regret which is a specialisation of Phi-equilibria when the set of deviations Phi is the set of all linear transformation from the set (mixed) strategy to itself. The authors provide efficient implementation of such no-linear-swap regret algorithms and prove that the set of linear-deviation correlated equilibria is a subset of extensive form correlated equilibria. The technical tool used is the of of Gordon 2008 which converts a regret minimiser for the space of deviations to a regret minimiser for the strategy set.

Strengths

The paper aims to solve an important problem, which is the complexity of CE in EFG. The paper is easy to read, clear and well written and the results seems non-trivial

Weaknesses

The notion of LCE is not well motivated. The authors prove that LCE are EFCE but, while EFCE are well motivated in EFG, LCE do not seems to have any practical meaning. Also the algorithm presented in the paper is slower at converging to EFCEs then the one specifically designed for them (Figure 2 left) and thus do not contribuite in this regard. The main weakness is that LCE are not an interesting equilibrium to converge to, except from their connection to EFCEs. The author should motivate the introduction of this new equilibria for example by finding examples in which LCE are a reasonable solution concept while other EFCEs are not. Without this the point the paper seems a technical exercise.

Questions

1) Th 31 seems to be the main technical statement of the paper. What are the main challenges of proving this statement? A better intuition of its implications/design would be appriciated 2) Motivate LCE (see Weaknesses) 3) Figure 2 only proves that the inclusions of CE, LCE and EFCE are strict, but is somewhat uninteresting once we have a theorem stating it. What happens is in Figure 2 left you put there time instead of iterations? If the no-linear-swap regret algorithm runs faster then the one that minimises only the trigger-regret this would be a better algorithm for finding EFCEs.

Rating

6: Weak Accept: Technically solid, moderate-to-high impact paper, with no major concerns with respect to evaluation, resources, reproducibility, ethical considerations.

Confidence

4: You are confident in your assessment, but not absolutely certain. It is unlikely, but not impossible, that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work.

Soundness

3 good

Presentation

3 good

Contribution

2 fair

Limitations

yes

Reviewer vZdR6/10 · confidence 2/52023-07-05

Summary

The paper studies regret minimization in extensive-form games (EFG). Specifically, they study a notion of regret called linear-swap regret, which measures the regret against linear transformations of the player's sequence-form strategies. This notion is stronger than trigger regret (as trigger deviations can be described as a linear transformation of sequence-form strategies) but weaker than swap regret. The main technical contribution lies in Theorem 3.1, which states that the set of linear transformations (from the set of sequence-form strategies to itself) is a compact polytope defined by a polynomial (in the game tree size) number of linear constraints. With that in hand, they leverage the template of Gordon et al. [2008] that produces a $\Phi$-regret minimization algorithm given a no-external-regret minimizer $\Phi$ and a fixed point oracle which returns fixed point of transformations in $\Phi$. In this context, the $\Phi$-regret minimization can be chosen, for example, to be Online Gradient Decent; and the fixed point oracle can be implemented using a Linear Programming Solver (e.g., Ellipsoid algorithm). This results in an efficient algorithm that achieves $\sqrt{T}$ linear-swap regret; which in turn also implies an efficient method for finding the correlated equilibrium that corresponds to linear transformations, that the authors call linear-deviation correlated equilibrium (LCE). Finally, the authors provide examples to show that the set of LCEs strictly contain**s** the set of CEs, and is strictly contain**ed** in the set of Extensive-form correlated equilibrium (EFCE). They also show that the set of Behavioral correlated equilibrium (BCE) and the set of LCEs are incomparable.

Strengths

The paper is written well; the authors give great background on EFGs, on different notions of correlated equilibrium, and their relation to different notions of regret. The construction of the algorithm and results are fairly easy to understand, and the authors express well the difference between their results and previous work by providing examples that compare their notion of equilibrium with existing notions. Finally, I agree that determining *"what is the strongest notion of rationality that can be attained efficiently"* is an interesting open problem, and indeed the paper provide some progress in this direction.

Weaknesses

- There is no obvious algorithmic contribution here: As noted, Theorem 3.1 is the main technical contribution which is purely geometric. Given that, constructing the algorithm is fairly straightforward with the scheme of Gordon et al. [2008]. - While the background the paper gives is quite solid, it is not very proportional to the relatively small amount of attention the authors give to their technical contribution. In particular, while the statement of Theorem 3.1 is clear, it is not very intuitive, and it is hard to understand why it is correct. This is also the main reason that my confidence score is relatively low. I think that some of the preliminaries, as well as the empirical evaluations, could be deferred to the appendix - instead, I think that giving the main proof ideas is more important. **Missing related work:** I think that the work of Anagnostides et al. (2022) is very relevant here and motivates important questions for future work. Specifically, this work achieves a much better rate of $\log T$ for trigger regret. Thus, it is fairly natural to ask whether this is also attainable for linear-swap regret, especially since they also built upon the template of Gordon et al. [2008]. Anagnostides, I., Farina, G., & Sandholm, T. (2022). Near-Optimal $\Phi $-Regret Learning in Extensive-Form Games. arXiv preprint arXiv:2208.09747.‏

Questions

N/A

Rating

6: Weak Accept: Technically solid, moderate-to-high impact paper, with no major concerns with respect to evaluation, resources, reproducibility, ethical considerations.

Confidence

2: You are willing to defend your assessment, but it is quite likely that you did not understand the central parts of the submission or that you are unfamiliar with some pieces of related work. Math/other details were not carefully checked.

Soundness

3 good

Presentation

3 good

Contribution

3 good

Limitations

N/A

Reviewer V8FX6/10 · confidence 4/52023-07-14

Summary

This paper focuses on addressing the challenge of minimizing linear swap regret in extensive form games, which is considered a stronger notion compared to trigger regret in Extensive Form Correlated Equilibrium (EFCE). To achieve efficient implementation of the Phi-regret minimization problem (where Phi set contains all linear swap transformation), the authors provide a polynomial-sized description of all linear swap operators. This result shows that there exists a no linear swap regret algorithm with polynomial-time iteration complexity based on the size of the game tree. Additionally, the paper presents experimental results that illustrate how the equilibria achieved through the no linear swap regret algorithm exhibit greater strength compared to EFCE.

Strengths

This paper introduces a novel and essential contribution by providing a polynomial characterization of the set of all linear transformations from a sequence-form strategy polytope to itself. This characterization holds significant value for future research on linear-swap-regret.

Weaknesses

The contribution of this paper may be perceived as limited. The primary and significant contribution lies in the polynomial description of the linear swap transformation set. Given the sequential structure of the sequence form strategy, this finding is not entirely unexpected. It appears that the paper lacks some intuitive explanations regarding the concept of linear swap (further details can be found in the questions raised). There is a sense that LCE (Linear Swap Correlated Equilibrium) is merely one equilibrium between CE (Correlated Equilibrium) and EFCE (Extensive Form Correlated Equilibrium), which is relatively easy to compute. The experimental results may not comprehensively demonstrate the potential and strength of LCE

Questions

1. Is there any intuition behind the linear swap transformation, apart from its linearity? For instance, can LCE be interpreted with the presence of a mediator? Why do you believe LCE is an equilibrium worth considering? 2. Maximizing social welfare is known to be challenging. However, in your experiments, did you observe any significant differences in social welfare between No-linear-swap-regret dynamics and No-trigger-regret dynamics? 3. Since No-linear-swap-regret dynamics are stronger than No-trigger-regret dynamics. The result in Figure 2 is expected. What about the running times? Is the running time of no-linear-swap-regret dynamics substantially larger? [1] Zhang, Brian, and Tuomas Sandholm. "Polynomial-time optimal equilibria with a mediator in extensive-form games." Advances in Neural Information Processing Systems 35 (2022): 24851-24863.

Rating

6: Weak Accept: Technically solid, moderate-to-high impact paper, with no major concerns with respect to evaluation, resources, reproducibility, ethical considerations.

Confidence

4: You are confident in your assessment, but not absolutely certain. It is unlikely, but not impossible, that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work.

Soundness

3 good

Presentation

3 good

Contribution

3 good

Limitations

The authors have partially addressed the limitations of their work, though there is space for improvement (see the section Weaknesses).

Reviewer vZdR2023-08-13

I thank the authors for their response - I have no further questions at this point.

Reviewer qFDY2023-08-15

I thank the authors for taking the time of answering my questions. I now understand better the questions that the paper tires to answer. However I'm confused about the arguments regarding [1]. Are the authors claiming that their algorithm can minimise linear swap regret also in bayesian games analysed in [1]? Moreover I still believe that while a general analysis of LCE can be deferred to future works, I think that a brief analysis of their features would greatly improve the paper. In particular a toy example in which EFCEs and LCEs leads to substantially different behaviour would be enough. I now increase my score. Depending on the authors answers I'll be willing to increase further my valuation. [1] Yishay Mansour, Mehryar Mohri, Jon Schneider, and Balasubramanian Sivan, 2022. Strategizing against learners in bayesian games. COLT.

Authorsrebuttal2023-08-18

Thank you a lot for engaging in this discussion and for providing continuous feedback on our paper. Indeed, our algorithm can minimize linear-swap regret in Bayesian games with a finite set of player types $\Theta$. This follows from the fact that extensive-form games are general enough to capture Bayesian games as well. In particular, we can represent any Bayesian game by using a game tree whose root is a chance node, which randomly initializes the player type to be one of the available $|\Theta|$ types. Information sets are set up so that each player can distinguish only their type and not the other players’. We have also included a short discussion of this in lines 118-123 in the Introduction of our submitted paper pdf. Regarding the brief analysis of the features of the LCE, we believe that the Example E.1 from the appendix, used to prove that LCE $\neq$ EFCE, is an interesting toy example showcasing the difference in behaviors that can be exhibited by these equilibria. In this case it happens that the linear deviations capture all possible deviations and, thus, LCE = CE. The game given in the example is a classic instance of a signaling game that has been extensively discussed in the past (eg. see von Stengel and Forges [2008] for an extensive discussion of the different behaviors that are possible for the EFCE and CE, which in our case is equal to the LCE). We will incorporate this discussion in the final version. [1] B. von Stengel and F. Forges. Extensive-form correlated equilibrium: Definition and computational complexity. Mathematics of Operations Research, 2008.

Reviewer Dcee2023-08-16

Thank you for further enlightening on the topic! Especially for the reference to the new recent work, that is interesting.

Program Chairsdecision2023-09-21

Decision

Accept (poster)

© 2026 NYSGPT2525 LLC