arxiv: 2604.02807 · v1 · submitted 2026-04-03 · 💻 cs.GT · math.OC

Recognition: no theorem link

Deception Equilibrium Analysis for Three-Party Stackelberg Game with Insider

Xiaoyu Xin , Gehui Xu , Yiguang Hong

Authors on Pith no claims yet

Pith reviewed 2026-05-13 18:31 UTC · model grok-4.3

classification 💻 cs.GT math.OC

keywords three-party gamedeception equilibriumStackelberg gamehyper Nash equilibriumsecurity gameinsider threathypergradient algorithmfalse data injection

0 comments

The pith

A three-party deception game framework lets the defender manipulate perceived parameters so equilibria stay consistent and utility invariant when move order changes.

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

The paper builds a unified leader-follower model for three parties—a defender, an insider, and external attackers—where the defender alters the game parameters that followers see. It defines Deception Stackelberg equilibria (DSE) for the hierarchical case and Hyper Nash equilibria (HNE) for simultaneous moves, then supplies necessary and sufficient conditions that make these two coincide. Under those conditions the defender’s payoff does not change when the game structure collapses from sequential to simultaneous. A hypergradient algorithm is given to locate the equilibria despite non-smooth best-response mappings, together with convergence guarantees. The same apparatus is applied to secure wireless links and to defense against insider-assisted false-data-injection attacks.

Core claim

We propose a unified three party leader follower game framework and introduce the concepts of Deception Stackelberg equilibria (DSE) and Hyper Nash equilibria (HNE), which generalize classical two-player Stackelberg and deception games. We develop necessary and sufficient conditions for the consistency between DSE and HNE, ensuring that the defender's utility remains invariant when the hierarchical structure degenerates into a simultaneous-move scenario. Moreover, we propose a scalable hypergradient-based algorithm with established convergence guarantees for seeking DSE.

What carries the argument

Deception Stackelberg equilibria (DSE) paired with Hyper Nash equilibria (HNE) under consistency conditions that keep defender utility unchanged when the leader-follower hierarchy flattens to simultaneous play.

If this is right

When the consistency conditions hold, the defender’s payoff is identical whether followers move sequentially or simultaneously.
The hypergradient method converges to DSE even though best-response maps are non-smooth and set-valued.
The same conditions guarantee robustness whenever followers’ observations remain uncertain or biased.
The framework directly yields improved defense strategies for wireless networks and for insider-assisted false-data-injection attacks.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The equivalence between DSE and HNE may let analysts replace a hard sequential game with an easier simultaneous one without loss of accuracy.
The same manipulation-of-perception idea could be tested in multi-agent economic or biological models that already contain information asymmetry.
Numerical checks on larger player counts would show whether the hypergradient scheme remains practical beyond the paper’s examples.

Load-bearing premise

Followers decide on the basis of misperceived and uncertain observations of the game parameters that the defender chooses.

What would settle it

A concrete three-party instance in which the stated necessary and sufficient consistency conditions are violated yet the defender’s utility is still identical under DSE and HNE would disprove the claimed equivalence.

Figures

Figures reproduced from arXiv: 2604.02807 by Gehui Xu, Xiaoyu Xin, Yiguang Hong.

**Figure 1.** Figure 1: The leader’s utility When the tuple (x ∗ , y∗ , z ∗ , θ∗ ) is a WDSE strategy, if f2(x ∗ , θ∗ ) ̸= 0, then the proof is complete. If f2(x ∗ , θ∗ ) = 0, then BRY (x, θ) = Ωy, thus, inf y∈Ωy UX(x ∗ , y,BRZ(x ∗ , y, θ∗ )) ≥ lim sup x→x∗ U¯X(x, θ∗ ). Then U¯X(x, θ∗ ) is upper semicontinuous at x ∗ . Consider a tuple (x0, y0, z0, θ0) and (x0, θ0) be a limit point of a ϵ-WDSE sequence as ϵ → 0. When f2(x0, θ0) ̸… view at source ↗

**Figure 2.** Figure 2: Robustness Assurance where the followers’ strategies y k+1 and z k+1 are obtained by Algorithm 2 given fixed x k . To determine the optimal strategy x ∗ (θ) for the fixed θ, we compare the utilities at the limiting solutions of these intervals with the leader’s utility at the zero points of f2. Specifically, at any zero point xi for i ∈ {1, . . . , xqθ }, the leader’s utility is evaluated based on the spec… view at source ↗

**Figure 3.** Figure 3: The leader’s utility. many iterates x k generated by Algorithm 1, then {α k e k}k∈N, with e k defined in (31), is summable. Let us analyze the convergence of Algorithm 1, whose proof is provided in Appendix B Theorem 4.1: Let Assumptions 2.1–2.3 and 4.1–4.2 hold. Suppose {α k , σk}k∈N satisfy the conditions in Lemma 4.5, and P(x k ) is a singleton for all but finitely many iterates on any interval Ωx,i. Th… view at source ↗

**Figure 5.** Figure 5: The Impact of Deception In [PITH_FULL_IMAGE:figures/full_fig_p010_5.png] view at source ↗

**Figure 6.** Figure 6: Convergence performance of Alg. 1 [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗

**Figure 7.** Figure 7: The relationship between WDSE and HNE under [PITH_FULL_IMAGE:figures/full_fig_p011_7.png] view at source ↗

**Figure 8.** Figure 8: Defense against IA-FDI attack in microgrid [PITH_FULL_IMAGE:figures/full_fig_p011_8.png] view at source ↗

**Figure 9.** Figure 9: The role of insider strategy space Ωx = [0, 5] into sub-intervals. Let Ω˜ x,1 = [0, 2.8], Ω˜ x,2 = [3.1, 5], and the gap interval I1 = [2.8, 3.1]. As shown in [PITH_FULL_IMAGE:figures/full_fig_p012_9.png] view at source ↗

**Figure 10.** Figure 10: ϵ-WDSE convergence [PITH_FULL_IMAGE:figures/full_fig_p012_10.png] view at source ↗

**Figure 11.** Figure 11: The relation between ϵ and gap interval I1 Setting Θ = {0.8, 1, 1.2}, we obtain θ ∗ = 0.8 [PITH_FULL_IMAGE:figures/full_fig_p012_11.png] view at source ↗

read the original abstract

This paper investigates strategic interactions within a three party deception security game involving a defender, an insider, and external attackers. We propose a robust deception mechanism where the leader manipulates game parameters perceived by followers to enhance defense performance when followers operate under misperceived and uncertain observation. Specifically, we propose a unified three party leader follower game framework and introduce the concepts of Deception Stackelberg equilibria (DSE) and Hyper Nash equilibria (HNE), which generalize classical two-player Stackelberg and deception games. We develop necessary and sufficient conditions for the consistency between DSE and HNE, ensuring that the defender's utility remains invariant when the hierarchical structure degenerates into a simultaneous-move scenario. Moreover, we propose a scalable hypergradient-based algorithm with established convergence guarantees for seeking DSE, efficiently addressing the computational challenges posed by non-smooth and set-valued best-response mappings. Finally, we apply theoretical analysis to practical scenarios in secure wireless communication and defense against insider-assisted false data injection attacks.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper defines DSE and HNE for three-party Stackelberg deception games with an insider and claims consistency conditions plus a hypergradient solver, though the set-valued responses may undermine the invariance claim.

read the letter

The key takeaway is that this work extends Stackelberg deception games to three parties including an insider who misperceives the game, defines DSE and HNE, and provides conditions under which the defender's utility stays the same if the game turns simultaneous. It also gives a hypergradient algorithm to find the DSE despite the non-smooth set-valued best responses. This is new in moving beyond two-player setups to handle an insider who operates with misperceived parameters. The consistency conditions and the algorithm for non-smooth set-valued best responses extend the classical literature in a structured way. The applications to secure wireless networks and defense against insider-assisted false data injection show where this could matter. The framework is presented clearly enough in the abstract, and the focus on computational challenges from set-valued mappings is a practical plus. If the full paper includes the proofs and some numerical results, it would strengthen the case. The potential weak point is exactly the one in the stress-test: the best-response mappings are non-smooth and set-valued, so the invariance under degeneration to simultaneous moves may not follow directly from the DSE definition without additional regularity conditions. Standard fixed-point arguments don't always work in set-valued cases, and if the paper relies on an implicit single-valued selection or overlooks the multiplicity from misperception, the necessity and sufficiency might not hold in full generality. It would help to see explicit examples or counterexamples to test this. Overall, this is aimed at game theorists and security researchers who model hierarchical interactions with deception. Someone looking for new tools in multi-party security games could find the definitions and algorithm useful, even if they need to verify the conditions themselves. I would send it to peer review. The ideas are fresh and the problem relevant, so referees can help tighten the proofs and check the algorithm's guarantees.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes a unified three-party leader-follower framework for deception security games with a defender, insider, and external attackers. It introduces Deception Stackelberg equilibria (DSE) and Hyper Nash equilibria (HNE) as generalizations of classical Stackelberg and Nash concepts, derives necessary and sufficient conditions for DSE-HNE consistency that ensure defender utility invariance when the hierarchical game degenerates to simultaneous moves, presents a hypergradient-based algorithm with convergence guarantees to compute DSE despite non-smooth set-valued best-response mappings, and applies the framework to secure wireless communication and insider-assisted false data injection attacks.

Significance. If the consistency conditions and convergence guarantees hold, the work extends two-player Stackelberg and deception games to a three-party setting with misperception, providing a new analytical tool for hierarchical security interactions. The algorithm's explicit handling of set-valued mappings and the practical applications in wireless security and FDI defense are strengths; the manuscript ships a computational method with stated theoretical guarantees.

major comments (2)

[Abstract / consistency theorem] Abstract and consistency conditions section: the necessary and sufficient conditions for DSE-HNE consistency are asserted to ensure defender utility invariance under degeneration to simultaneous moves, yet the abstract explicitly states that best-response mappings are non-smooth and set-valued. Standard fixed-point arguments for such equivalence require additional regularity (upper hemicontinuity plus convex-valuedness or a selection rule); without these being stated and verified, the claimed necessity and sufficiency do not automatically extend to the general case.
[Algorithm and convergence analysis] Algorithm section: the hypergradient-based method claims convergence despite set-valued best responses, but the proof sketch must specify how the hypergradient is defined or approximated when the insider's misperception induces multiplicity; otherwise the convergence guarantee is not load-bearing for the stated generality.

minor comments (2)

[Introduction] Introduction: the distinction between DSE and classical Stackelberg equilibrium could be clarified with a short side-by-side comparison table of equilibrium definitions.
[Preliminaries] Notation: ensure that the symbols for perceived vs. true game parameters are introduced before their first use in the equilibrium definitions.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed comments. We have revised the manuscript to explicitly address the regularity conditions for the consistency theorem and to clarify the definition and selection of the hypergradient in the presence of set-valued best responses. Point-by-point responses follow.

read point-by-point responses

Referee: [Abstract / consistency theorem] Abstract and consistency conditions section: the necessary and sufficient conditions for DSE-HNE consistency are asserted to ensure defender utility invariance under degeneration to simultaneous moves, yet the abstract explicitly states that best-response mappings are non-smooth and set-valued. Standard fixed-point arguments for such equivalence require additional regularity (upper hemicontinuity plus convex-valuedness or a selection rule); without these being stated and verified, the claimed necessity and sufficiency do not automatically extend to the general case.

Authors: We thank the referee for highlighting this important technical point. In our model, the best-response mappings are upper hemicontinuous and convex-valued because the followers' utility functions are continuous and the strategy sets are compact and convex; these properties are preserved under the linear misperception structure induced by the insider. In the revised manuscript we have added an explicit lemma (Lemma 3.1) verifying upper hemicontinuity and convex-valuedness, and we have restated Theorem 3.2 to invoke these conditions directly in the fixed-point argument. The necessity and sufficiency of the DSE-HNE consistency conditions therefore continue to hold. We have also updated the abstract to note that the claimed invariance relies on these regularity properties. revision: yes
Referee: [Algorithm and convergence analysis] Algorithm section: the hypergradient-based method claims convergence despite set-valued best responses, but the proof sketch must specify how the hypergradient is defined or approximated when the insider's misperception induces multiplicity; otherwise the convergence guarantee is not load-bearing for the stated generality.

Authors: We agree that the original sketch required additional detail. In the revised version we define the hypergradient via the minimal-norm element of the convex hull of the Clarke subdifferential of the leader's utility when the insider's best-response set is multi-valued. Because the subdifferential is nonempty, convex, and compact under our continuity and compactness assumptions, this selection is always well-defined and unique. The convergence proof in the appendix has been expanded to show that the resulting hypergradient descent sequence converges to a DSE at a rate governed by the Lipschitz constant of the composed utility, thereby supporting the stated generality. revision: yes

Circularity Check

0 steps flagged

No circularity: DSE/HNE consistency derived from standard generalizations

full rationale

The paper defines Deception Stackelberg equilibria (DSE) and Hyper Nash equilibria (HNE) as explicit generalizations of classical two-player Stackelberg and deception games, then states necessary and sufficient conditions for their consistency (defender utility invariance under degeneration to simultaneous moves). No quoted equations or text reduce any claimed prediction or condition to a fitted parameter, self-referential definition, or load-bearing self-citation. The non-smooth set-valued best-response mappings are acknowledged as a computational challenge addressed by a separate hypergradient algorithm with stated convergence guarantees. The derivation chain therefore remains self-contained against external game-theoretic benchmarks and does not collapse by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 2 invented entities

The central claim rests on the domain assumption that followers have misperceived and uncertain observations, allowing parameter manipulation by the leader; no free parameters or invented entities beyond the new equilibrium concepts are specified in the abstract.

axioms (1)

domain assumption Followers operate under misperceived and uncertain observation
Explicitly stated as the basis for the robust deception mechanism in the abstract.

invented entities (2)

Deception Stackelberg equilibria (DSE) no independent evidence
purpose: Equilibrium concept capturing leader's manipulation of perceived parameters in three-party game
Newly introduced to generalize classical Stackelberg games to deception settings.
Hyper Nash equilibria (HNE) no independent evidence
purpose: Equilibrium concept for simultaneous-move version with deception
Newly introduced to analyze consistency with DSE.

pith-pipeline@v0.9.0 · 5471 in / 1277 out tokens · 43084 ms · 2026-05-13T18:31:26.627162+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

43 extracted references · 43 canonical work pages

[1]

Three-stage stackelberg game for defending against full-duplex active eavesdropping attacks in cooperative communication,

H. Fang, L. Xu, Y . Zou, X. Wang, and K.-K. R. Choo, “Three-stage stackelberg game for defending against full-duplex active eavesdropping attacks in cooperative communication,”IEEE Transactions on Vehicular Technology, vol. 67, no. 11, pp. 10 788–10 799, 2018

work page 2018
[2]

Combating full-duplex active eavesdropper: A hierarchical game perspective,

X. Tang, P. Ren, Y . Wang, and Z. Han, “Combating full-duplex active eavesdropper: A hierarchical game perspective,”IEEE Transactions on Communications, vol. 65, no. 3, pp. 1379–1395, 2017

work page 2017
[3]

An efficient cryptographic scheme for securing time-sensitive micro- grid communications under key leakage and dishonest insiders,

J. Wang, J. Appiah-Kubi, L.-A. Lee, D. Shi, D. Zou, and C.-C. Liu, “An efficient cryptographic scheme for securing time-sensitive micro- grid communications under key leakage and dishonest insiders,”IEEE Transactions on Smart Grid, vol. 14, no. 2, pp. 1210–1222, 2023

work page 2023
[4]

False data injection attacks and the insider threat in smart systems,

S. G ¨onen, H. H. Sayan, E. N. Yılmaz, F.¨Ust¨unsoy, and G. Karacayılmaz, “False data injection attacks and the insider threat in smart systems,” Computers & Security, vol. 97, p. 101955, 2020. 16

work page 2020
[5]

Collaborative honeypot defense in uav networks: A learning-based game approach,

Y . Wang, Z. Su, A. Benslimane, Q. Xu, M. Dai, and R. Li, “Collaborative honeypot defense in uav networks: A learning-based game approach,” IEEE Transactions on Information Forensics and Security, vol. 19, pp. 1963–1978, 2024

work page 1963
[6]

2025 cost of insider risks global report,

DTEX, “2025 cost of insider risks global report,” 2025. [Online]. Available: https://dtexsystems.com

work page 2025
[7]

D. M. Cappelli, A. P. Moore, and R. F. Trzeciak,The CERT Guide to Insider Threats: How to Prevent, Detect, and Respond to Information Technology Crimes (Theft, Sabotage, Fraud). Addison-Wesley, 2012

work page 2012
[8]

Consistency of Stackelberg and Nash equilibria in three-player leader-follower games,

G. Xu, G. Chen, Z. Cheng, Y . Hong, and H. Qi, “Consistency of Stackelberg and Nash equilibria in three-player leader-follower games,” IEEE Transactions on Information Forensics and Security, vol. 19, pp. 5330–5344, 2024

work page 2024
[9]

Deception techniques in computer security: A research perspective,

X. Han, N. Kheir, and D. Balzarotti, “Deception techniques in computer security: A research perspective,”ACM Comput. Surv., vol. 51, no. 4, Jul. 2018

work page 2018
[10]

Stackelberg game based relay selection for physical layer security and energy efficiency enhancement in cognitive radio networks,

H. Fang, L. Xu, and K.-K. R. Choo, “Stackelberg game based relay selection for physical layer security and energy efficiency enhancement in cognitive radio networks,”Applied Mathematics and Computation, vol. 296, pp. 153–167, 2017

work page 2017
[11]

Defense strategy against load redistribution attacks on power systems considering insider threats,

Z. Liu and L. Wang, “Defense strategy against load redistribution attacks on power systems considering insider threats,”IEEE Transactions on Smart Grid, vol. 12, no. 2, pp. 1529–1540, 2021

work page 2021
[12]

Single-leader-multiple-followers Stackelberg security game with hypergame framework,

Z. Cheng, G. Chen, and Y . Hong, “Single-leader-multiple-followers Stackelberg security game with hypergame framework,”IEEE Trans- actions on Information Forensics and Security, vol. 17, p. 954–969, 2022

work page 2022
[13]

Imitative attacker deception in Stackelberg security games

T. Nguyen and H. Xu, “Imitative attacker deception in Stackelberg security games.” inIJCAI, 2019, pp. 528–534

work page 2019
[14]

Weak via strong Stackelberg problem: new results,

P. Loridan and J. Morgan, “Weak via strong Stackelberg problem: new results,”Journal of Global Optimization, vol. 8, no. 3, pp. 263–287, 1996

work page 1996
[15]

On the inducibility of stackelberg equilibrium for security games,

Q. Guo, J. Gan, F. Fang, L. Tran-Thanh, M. Tambe, and B. An, “On the inducibility of stackelberg equilibrium for security games,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 33, no. 01, 2019, pp. 2020–2028

work page 2019
[16]

Interdependent strategic security risk management with bounded rationality in the internet of things,

J. Chen and Q. Zhu, “Interdependent strategic security risk management with bounded rationality in the internet of things,”IEEE Transactions on Information Forensics and Security, vol. 14, no. 11, pp. 2958–2971, 2019

work page 2019
[17]

Security games with incomplete information,

K. C. Nguyen, T. Alpcan, and T. Basar, “Security games with incomplete information,” in2009 IEEE International Conference on Communica- tions. IEEE, 2009, pp. 1–6

work page 2009
[18]

Dynamic defense strategy against stealth malware propagation in cyber-physical systems,

K. Xiao, C. Zhu, J. Xie, Y . Zhou, X. Zhu, and W. Zhang, “Dynamic defense strategy against stealth malware propagation in cyber-physical systems,” inIEEE INFOCOM 2018 - IEEE Conference on Computer Communications, 2018, pp. 1790–1798

work page 2018
[19]

Anti-jamming transmission stack- elberg game with observation errors,

L. Xiao, T. Chen, J. Liu, and H. Dai, “Anti-jamming transmission stack- elberg game with observation errors,”IEEE Communications Letters, vol. 19, no. 6, pp. 949–952, 2015

work page 2015
[20]

Hypergame theory: a model for conflict, misperception, and deception,

N. S. Kovach, A. S. Gibson, and G. B. Lamont, “Hypergame theory: a model for conflict, misperception, and deception,”Game Theory, vol. 2015, no. 1, p. 570639, 2015

work page 2015
[21]

Non-cooperative games,

J. Nash, “Non-cooperative games,”Annals of Mathematics, vol. 54, no. 2, pp. 286–295, 1951

work page 1951
[22]

Preservation of misperceptions – stability analysis of hyper- games,

Y . Sasaki, “Preservation of misperceptions – stability analysis of hyper- games,”Proceedings of the 52nd Annual Meeting of the ISSS - 2008, Madison, Wisconsin, vol. 3, no. 1, Jul. 2008

work page 2008
[23]

A gradient method for multilevel optimization,

R. Sato, M. Tanaka, and A. Takeda, “A gradient method for multilevel optimization,” inAdvances in Neural Information Processing Systems, vol. 34. Curran Associates, Inc., 2021, pp. 7522–7533

work page 2021
[24]

Two-level value function approach to non-smooth optimistic and pessimistic bilevel programs,

S. Dempe, B. S. Mordukhovich, and A. B. Zemkoho, “Two-level value function approach to non-smooth optimistic and pessimistic bilevel programs,”Optimization, vol. 68, no. 2-3, pp. 433–455, 2019. [Online]. Available: https://doi.org/10.1080/02331934.2018.1543294

work page doi:10.1080/02331934.2018.1543294 2019
[25]

Secure transmission with guaranteed user satisfaction in Heterogeneous Networks: A two-level Stackelberg game approach,

N. Wu, X. Zhou, and M. Sun, “Secure transmission with guaranteed user satisfaction in Heterogeneous Networks: A two-level Stackelberg game approach,”IEEE Transactions on Communications, vol. 66, no. 6, pp. 2738–2750, 2018

work page 2018
[26]

A three-stage stackelberg game for secure communica- tion with a wireless powered jammer,

Q. Li and D. Xu, “A three-stage stackelberg game for secure communica- tion with a wireless powered jammer,” in2019 11th International Con- ference on Wireless Communications and Signal Processing (WCSP), 2019, pp. 1–6

work page 2019
[27]

A bayesian game-enhanced auction model for federated cloud services using blockchain,

Z. Shi, H. Zhou, C. de Laat, and Z. Zhao, “A bayesian game-enhanced auction model for federated cloud services using blockchain,”Future Generation Computer Systems, vol. 136, pp. 49–66, 2022

work page 2022
[28]

First-level hypergame for investigating two decision-maker conflicts with unknown misperceptions of preferences within the framework of gmcr,

S. Zhao, L. Xue, B. A. Addae, J. Wu, and D. Wang, “First-level hypergame for investigating two decision-maker conflicts with unknown misperceptions of preferences within the framework of gmcr,”Expert Systems with Applications, vol. 237, p. 121619, 2024

work page 2024
[29]

Implicit learning dynamics in stackelberg games: equilibria characterization, convergence analysis, and empirical study,

T. Fiez, B. Chasnov, and L. Ratliff, “Implicit learning dynamics in stackelberg games: equilibria characterization, convergence analysis, and empirical study,” inProceedings of the 37th International Conference on Machine Learning, ser. ICML’20. JMLR.org, 2020

work page 2020
[30]

General sum markov games for strategic detection of advanced persistent threats using moving target defense in cloud networks,

S. Sengupta, A. Chowdhary, D. Huang, and S. Kambhampati, “General sum markov games for strategic detection of advanced persistent threats using moving target defense in cloud networks,” inInternational Con- ference on Decision and Game Theory for Security. Cham: Springer International Publishing, 2019, pp. 492–512

work page 2019
[31]

Big hype: Best intervention in games via distributed hy- pergradient descent,

P. D. Grontas, G. Belgioioso, C. Cenedese, M. Fochesato, J. Lygeros, and F. D¨orfler, “Big hype: Best intervention in games via distributed hy- pergradient descent,”IEEE Transactions on Automatic Control, vol. 69, no. 12, pp. 8338–8353, 2024

work page 2024
[32]

Convergence properties of a regularization scheme for mathematical programs with complementarity constraints,

S. Scholtes, “Convergence properties of a regularization scheme for mathematical programs with complementarity constraints,”SIAM Jour- nal on Optimization, vol. 11, no. 4, pp. 918–936, 2001

work page 2001
[33]

Conservative set valued fields, automatic differentiation, stochastic gradient methods and deep learning,

J. Bolte and E. Pauwels, “Conservative set valued fields, automatic differentiation, stochastic gradient methods and deep learning,”Mathe- matical Programming, vol. 188, pp. 19–51, 2021

work page 2021
[34]

A game-theoretic taxonomy and survey of defensive deception for cybersecurity and privacy,

J. Pawlick, E. Colbert, and Q. Zhu, “A game-theoretic taxonomy and survey of defensive deception for cybersecurity and privacy,”ACM Computing Surveys (CSUR), vol. 52, no. 4, pp. 1–28, 2019

work page 2019
[35]

Stealthy attacks meets insider threats: A three-player game model,

X. Feng, Z. Zheng, P. Hu, D. Cansever, and P. Mohapatra, “Stealthy attacks meets insider threats: A three-player game model,” inIEEE Military Communications Conference, 2015, pp. 25–30

work page 2015
[36]

Social media and misleading information in a democracy: A mechanism design approach,

A. Dave, I. V . Chremos, and A. A. Malikopoulos, “Social media and misleading information in a democracy: A mechanism design approach,” IEEE Transactions on Automatic Control, vol. 67, no. 5, pp. 2633–2639, 2022

work page 2022
[37]

Bas ¸ar and G

T. Bas ¸ar and G. J. Olsder,Dynamic Noncooperative Game Theory, 2nd Edition. Society for Industrial and Applied Mathematics, 1998

work page 1998
[38]

Sample-efficient learning of stackelberg equilibria in general-sum games,

Y . Bai, C. Jin, H. Wang, and C. Xiong, “Sample-efficient learning of stackelberg equilibria in general-sum games,”Advances in Neural Information Processing Systems, vol. 34, pp. 25 799–25 811, 2021

work page 2021
[39]

Facchinei and J.-S

F. Facchinei and J.-S. Pang,Finite-dimensional Variational Inequalities and Complementarity Problems. Springer, 2003

work page 2003
[40]

Nonsmooth implicit differentiation for machine-learning and optimization,

J. Bolte, T. Le, E. Pauwels, and T. Silveti-Falls, “Nonsmooth implicit differentiation for machine-learning and optimization,”Advances in Neural Information Processing Systems, vol. 34, pp. 13 537–13 549, 2021

work page 2021
[41]

Geometric categories and o-minimal structures,

L. Van den Dries and C. Miller, “Geometric categories and o-minimal structures,” 1996

work page 1996
[42]

Stochastic subgradient method converges on tame functions,

D. Davis, D. Drusvyatskiy, S. Kakade, and J. D. Lee, “Stochastic subgradient method converges on tame functions,”Foundations of Com- putational Mathematics, vol. 20, no. 1, pp. 119–154, 2020

work page 2020
[43]

Coordinated multiple-relays based physical-layer security improvement: A single-leader multiple-followers Stackelberg game scheme,

H. Fang, L. Xu, and X. Wang, “Coordinated multiple-relays based physical-layer security improvement: A single-leader multiple-followers Stackelberg game scheme,”IEEE Transactions on Information Forensics and Security, vol. 13, no. 1, pp. 197–209, 2018

work page 2018