arXiv subjects

Anders Jonsson

Publications and source records attributed to Anders Jonsson.

At least 19 recordsLinked to original sources

arXiv2026

Sample Efficient Hierarchical Reinforcement Learning via Best Policy Identification

We present HBPI-UCRL, a model-based algorithm for hierarchical reinforcement learning (HRL) that learns high-level and low-level policies in parallel. HBPI-UCRL exploits the fact that a high-level transition corresponds to a multi-step transition at the low level. We introduce two conditions on the low-level dynamics that are sufficient to make parallel HRL learnable. When these conditions hold, we prove that HBPI-UCRL has a polynomial sample complexity in the problem parameters. In the sparse-reward, goal-directed setting, our sample complexity upper bound for HBPI-UCRL is strictly lower than that of its non-hierarchical counterpart, providing theoretical justification for the empirical success of HRL.

cs.LG↗

arXiv2026

Coordination Graphs for Constrained Multi-Agent Reinforcement Learning

Constrained Multi-agent reinforcement learning (CMARL) faces two intertwined challenges: the joint action space grows exponentially with the number of agents, and additional requirements couple agents in ways that reward structure alone does not capture. We introduce Coordination Graphs for Constrained Multi-Agent Reinforcement Learning (CG-CMARL), a framework that addresses both challenges by combining coordination graphs with Lagrangian duality. The system decomposes the joint problem into pairwise regions, each served by a set of shared Q-functions, one for the primary objective and one for each of the constraints, so that the number of learned models is independent of the number of agents. At execution time, Max-Sum message passing coordinates actions across the factor graph, while a Lagrangian multiplier controls the objective--constraint tradeoff, allowing a single trained model to trace a Pareto front without retraining. We provide convergence guarantees under mild conditions, together with a compositional error bound that decomposes into separate interpretable sources, each traceable to a specific design choice and independently controllable. Experiments on cooperative navigation tasks (where teams of up to 10 agents must coordinate to reach target positions while satisfying pairwise constraints) show that our method produces Pareto fronts dominating established baselines trained at fixed reward-shaping ratios, while scaling to team sizes where centralized approaches become intractable.

cs.AI↗

arXiv2026

The Terminal Representation in Reinforcement Learning

Representation learning is a powerful tool for spatio-temporal abstraction within reinforcement learning (RL). Two well established approaches are through the successor representation (SR) and the default representation (DR). The SR encodes states by the future trajectories they induce, capturing information flow decoupled from reward. The DR builds on this by weighting trajectories with reward, integrating credit-assignment structure into the representation. Eigenvectors of both representations have been used to support a range of downstream tasks -- including option discovery, reward shaping, transfer learning, and exploration. We introduce a structurally distinct formulation: the terminal representation (TR). The TR encodes reward-weighted trajectories similarly to the DR, but can be learned as a lower-dimensionality object, and can be used directly for the mentioned applications without eigenvector computations. Eigendecomposition also imposes the assumption of symmetric transition dynamics, which the TR can bypass. In this work we develop the theoretical foundations of the TR: its derivation, convergence of two learning algorithms, its use for zero-shot compositionality, and equivalences between alternative reward formulations. We further show the TR is embedded in the top DR eigenvector, allowing it to capture the same underlying knowledge without eigendecomposition. Additionally, we provide empirical evidence of the TR as a viable alternative to existing representations in subsidiary applications, while requiring less computational overhead to learn, store, and use.

cs.LG↗

arXiv2026

Scalable Constrained Multi-Agent Reinforcement Learning via State Augmentation and Consensus for Separable Dynamics

We present a distributed approach for constrained Multi-Agent Reinforcement Learning (MARL) that combines state-augmented policy learning with distributed consensus over dual variables. Our method targets systems where agents have separable dynamics but must coordinate to satisfy global resource constraints, a setting in which, as we demonstrate empirically, independent learning fails to produce feasible solutions because agents cannot determine appropriate individual contributions toward collective constraint satisfaction. The key technical contribution is showing that lightweight neighbor-to-neighbor consensus over Lagrange multipliers suffices for globally coordinated constraint enforcement while preserving the scalability of independent training. Each agent learns a single augmented policy offline, conditioned on both its local state and a dual variable encoding constraint feedback. During execution, agents reach agreement on this dual variable through local communication alone. We prove that under mild connectivity assumptions, the consensus error among agents' multipliers is bounded, and show that this translates to a bounded constraint violation that decreases with graph connectivity and the number of consensus rounds. Unlike centralized training with decentralized execution (CTDE) approaches, whose complexity grows at least quadratically with agent count, our method scales linearly in both training and execution. Experiments on smart grid demand response demonstrate that consensus coordination is \emph{essential for feasibility}: without it, agents satisfy grid capacity constraints only by indefinitely postponing demand, a degenerate non-solution. With consensus, agents converge to a shared dual variable and satisfy both grid constraints and demand fulfillment, scaling to thousands of agents while CTDE baselines are limited to dozens.

cs.LG↗

arXiv2026

Sampling-guided exploration of active feature selection policies

Determining the most appropriate features for machine learning predictive models is challenging regarding performance and feature acquisition costs. In particular, global feature choice is limited given that some features will only benefit a subset of instances. In previous work, we proposed a reinforcement learning approach to sequentially recommend which modality to acquire next to reach the best information/cost ratio, based on the instance-specific information already acquired. We formulated the problem as a Markov Decision Process where the state's dimensionality changes during the episode, avoiding data imputation, contrary to existing works. However, this only allowed processing a small number of features, as all possible combinations of features were considered. Here, we address these limitations with two contributions: 1) we expand our framework to larger datasets with a heuristic-based strategy that focuses on the most promising feature combinations, and 2) we introduce a post-fit regularisation strategy that reduces the number of different feature combinations, leading to compact sequences of decisions. We tested our method on four binary classification datasets (one involving high-dimensional variables), the largest of which had 56 features and 4500 samples. We obtained better performance than state-of-the-art methods, both in terms of accuracy and policy complexity.

cs.LG↗

arXiv2026

Intermediate Results on the Complexity of STRIPS$_{1}^{1}$

This paper is based on Bylander's results on the computational complexity of propositional STRIPS planning. He showed that when only ground literals are permitted, determining plan existence is PSPACE-complete even if operators are limited to two preconditions and two postconditions. While NP-hardness is settled, it is unknown whether propositional STRIPS with operators that only have one precondition and one effect is NP-complete. We shed light on the question whether this small solution hypothesis for STRIPS$^1_1$ is true, calling a SAT solver for small instances, introducing the literal graph, and mapping it to Petri nets.

cs.AI↗

arXiv2025

Performance Evaluation of Multi-Armed Bandit Algorithms for Wi-Fi Channel Access

The adoption of dynamic, self-learning solutions for real-time wireless network optimization has recently gained significant attention due to the limited adaptability of existing protocols. This paper investigates multi-armed bandit (MAB) strategies as a data-driven approach for decentralized, online channel access optimization in Wi-Fi, targeting dynamic channel access settings: primary channel, channel width, and contention window (CW) adjustment. Key design aspects are examined, including the adoption of joint versus factorial action spaces, the inclusion of contextual information, and the nature of the action-selection strategy (optimism-driven, unimodal, or randomized). State-of-the-art algorithms and a proposed lightweight contextual approach, E-RLB, are evaluated through simulations. Results show that contextual and optimism-driven strategies consistently achieve the highest performance and fastest adaptation under recurrent conditions. Unimodal structures require careful graph construction to ensure that the unimodality assumption holds. Randomized exploration, adopted in the proposed E-RLB, can induce disruptive parameter reallocations, especially in multi-player settings. Decomposing the action space across several specialized agents accelerates convergence but increases sensitivity to randomized exploration and demands coordination under shared rewards to avoid correlated learning. Finally, despite its inherent inefficiencies from epsilon-greedy exploration, E-RLB demonstrates effective adaptation and learning, highlighting its potential as a viable low-complexity solution for realistic dynamic deployments.

cs.NI↗

arXiv2025

Learning-Based Channel Access in Wi-Fi: A Multi-Armed Bandit Approach

Due to its static protocol design, IEEE 802.11 (aka Wi-Fi) channel access lacks adaptability to address dynamic network conditions, resulting in inefficient spectrum utilization, unnecessary contention, and packet collisions. This paper investigates reinforcement learning (RL) solutions to optimize Wi-Fi's medium access control (MAC). In particular, a multi-armed bandit (MAB) framework is proposed for dynamic channel access (including both the primary channel and channel width) and contention window (CW) adjustment. In this setting, we study relevant learning design principles such as adopting joint or factorial action spaces (handled by a single agent (SA) and multiple agents (MA), respectively) and the importance of incorporating contextual information. Our simulation results show that cooperative MA architectures converge faster than their SA counterparts, as agents operate over smaller action spaces. Another key insight is that contextual MAB algorithms consistently outperform non-contextual ones, highlighting the value of leveraging side information in action selection. Moreover, in multi-player settings, results demonstrate that decentralized learners can achieve implicit coordination, although their greediness may degrade coexisting networks' performance and induce policy-chasing dynamics. Overall, these findings demonstrate that (contextual) MAB-based learning offers a practical and adaptive alternative to static IEEE 802.11 protocols, enabling more efficient and intelligent spectrum utilization.

cs.NI↗

arXiv2025

Learning The Minimum Action Distance

This paper presents a state representation framework for Markov decision processes (MDPs) that can be learned solely from state trajectories, requiring neither reward signals nor the actions executed by the agent. We propose learning the minimum action distance (MAD), defined as the minimum number of actions required to transition between states, as a fundamental metric that captures the underlying structure of an environment. MAD naturally enables critical downstream tasks such as goal-conditioned reinforcement learning and reward shaping by providing a dense, geometrically meaningful measure of progress. Our self-supervised learning approach constructs an embedding space where the distances between embedded state pairs correspond to their MAD, accommodating both symmetric and asymmetric approximations. We evaluate the framework on a comprehensive suite of environments with known MAD values, encompassing both deterministic and stochastic dynamics, as well as discrete and continuous state spaces, and environments with noisy observations. Empirical results demonstrate that the proposed approach not only efficiently learns accurate MAD representations across these diverse settings but also significantly outperforms existing state representation methods in terms of representation quality.

cs.LG↗

arXiv2025

Distances for Markov chains from sample streams

Bisimulation metrics are powerful tools for measuring similarities between stochastic processes, and specifically Markov chains. Recent advances have uncovered that bisimulation metrics are, in fact, optimal-transport distances, which has enabled the development of fast algorithms for computing such metrics with provable accuracy and runtime guarantees. However, these recent methods, as well as all previously known methods, assume full knowledge of the transition dynamics. This is often an impractical assumption in most real-world scenarios, where typically only sample trajectories are available. In this work, we propose a stochastic optimization method that addresses this limitation and estimates bisimulation metrics based on sample access, without requiring explicit transition models. Our approach is derived from a new linear programming (LP) formulation of bisimulation metrics, which we solve using a stochastic primal-dual optimization method. We provide theoretical guarantees on the sample complexity of the algorithm and validate its effectiveness through a series of empirical evaluations.

cs.LG↗

arXiv2025

Towards Specialized Wireless Networks Using an ML-Driven Radio Interface

Future wireless networks will need to support diverse applications (such as extended reality), scenarios (such as fully automated industries), and technological advances (such as terahertz communications). Current wireless networks are designed to perform adequately across multiple scenarios so they lack the adaptability needed for specific use cases. Therefore, meeting the stringent requirements of next-generation applications incorporating technology advances and operating in novel scenarios will necessitate wireless specialized networks which we refer to as SpecNets. These networks, equipped with cognitive capabilities, dynamically adapt to the unique demands of each application, e.g., by automatically selecting and configuring network mechanisms. An enabler of SpecNets are the recent advances in artificial intelligence and machine learning (AI/ML), which allow to continuously learn and react to changing requirements and scenarios. By integrating AI/ML functionalities, SpecNets will fully leverage the concept of AI/ML-defined radios (MLDRs) that are able to autonomously establish their own communication protocols by acquiring contextual information and dynamically adapting to it. In this paper, we introduce SpecNets and explain how MLDR interfaces enable this concept. We present three illustrative use cases for wireless local area networks (WLANs): bespoke industrial networks, traffic-aware robust THz links, and coexisting networks. Finally, we showcase SpecNets' benefits in the industrial use case by introducing a lightweight, fast-converging ML agent based on multi-armed bandits (MABs). This agent dynamically optimizes channel access to meet varying performance needs: high throughput, low delay, or fair access. Results demonstrate significant gains over IEEE 802.11, highlighting the system's autonomous adaptability across diverse scenarios.

cs.NI↗

arXiv2024

Provably Efficient Exploration in Reward Machines with Low Regret

We study reinforcement learning (RL) for decision processes with non-Markovian reward, in which high-level knowledge of the task in the form of reward machines is available to the learner. We consider probabilistic reward machines with initially unknown dynamics, and investigate RL under the average-reward criterion, where the learning performance is assessed through the notion of regret. Our main algorithmic contribution is a model-based RL algorithm for decision processes involving probabilistic reward machines that is capable of exploiting the structure induced by such machines. We further derive high-probability and non-asymptotic bounds on its regret and demonstrate the gain in terms of regret over existing algorithms that could be applied, but obliviously to the structure. We also present a regret lower bound for the studied setting. To the best of our knowledge, the proposed algorithm constitutes the first attempt to tailor and analyze regret specifically for RL with probabilistic reward machines.

cs.LG↗

arXiv2024

Tractable Offline Learning of Regular Decision Processes

This work studies offline Reinforcement Learning (RL) in a class of non-Markovian environments called Regular Decision Processes (RDPs). In RDPs, the unknown dependency of future observations and rewards from the past interactions can be captured by some hidden finite-state automaton. For this reason, many RDP algorithms first reconstruct this unknown dependency using automata learning techniques. In this paper, we show that it is possible to overcome two strong limitations of previous offline RL algorithms for RDPs, notably RegORL. This can be accomplished via the introduction of two original techniques: the development of a new pseudometric based on formal languages, which removes a problematic dependency on $L_\infty^\mathsf{p}$-distinguishability parameters, and the adoption of Count-Min-Sketch (CMS), instead of naive counting. The former reduces the number of samples required in environments that are characterized by a low complexity in language-theoretic terms. The latter alleviates the memory requirements for long planning horizons. We derive the PAC sample complexity bounds associated to each of these techniques, and we validate the approach experimentally.

cs.LG↗

arXiv2024

Hierarchical Average-Reward Linearly-solvable Markov Decision Processes

We introduce a novel approach to hierarchical reinforcement learning for Linearly-solvable Markov Decision Processes (LMDPs) in the infinite-horizon average-reward setting. Unlike previous work, our approach allows learning low-level and high-level tasks simultaneously, without imposing limiting restrictions on the low-level tasks. Our method relies on partitions of the state space that create smaller subtasks that are easier to solve, and the equivalence between such partitions to learn more efficiently. We then exploit the compositionality of low-level tasks to exactly represent the value function of the high-level task. Experiments show that our approach can outperform flat average-reward reinforcement learning by one or several orders of magnitude.

cs.LG↗

arXiv2024

Performance Evaluation of MLO for XR Streaming: Can Wi-Fi 7 Meet the Expectations?

Extended Reality (XR) has stringent throughput and delay requirements that are hard to meet with current wireless technologies. Missing these requirements can lead to worsened picture quality, perceived lag between user input and corresponding output, and even dizziness for the end user. In this paper, we study the capability of upcoming Wi-Fi 7, and its novel support for Multi-Link Operation (MLO), to cope with these tight requirements. Our study is based on simulation results extracted from an MLO-compliant simulator that realistically reproduces VR traffic. Results show that MLO can sustain VR applications. By jointly using multiple links with independent channel access procedures, MLO can reduce the overall delay, which is especially useful in the uplink, as it has more stringent requirements than the downlink, and is instrumental in delivering the expected performance. We show that using MLO can allow more users per network than an equivalent number of links using SLO. We also show that while maintaining the same overall bandwidth, a higher number of MLO links with narrow channels leads to lower delays than a lower number of links with wider channels.

cs.NI↗

arXiv2024

Bisimulation Metrics are Optimal Transport Distances, and Can be Computed Efficiently

We propose a new framework for formulating optimal transport distances between Markov chains. Previously known formulations studied couplings between the entire joint distribution induced by the chains, and derived solutions via a reduction to dynamic programming (DP) in an appropriately defined Markov decision process. This formulation has, however, not led to particularly efficient algorithms so far, since computing the associated DP operators requires fully solving a static optimal transport problem, and these operators need to be applied numerous times during the overall optimization process. In this work, we develop an alternative perspective by considering couplings between a flattened version of the joint distributions that we call discounted occupancy couplings, and show that calculating optimal transport distances in the full space of joint distributions can be equivalently formulated as solving a linear program (LP) in this reduced space. This LP formulation allows us to port several algorithmic ideas from other areas of optimal transport theory. In particular, our formulation makes it possible to introduce an appropriate notion of entropy regularization into the optimization problem, which in turn enables us to directly calculate optimal transport distances via a Sinkhorn-like method we call Sinkhorn Value Iteration (SVI). We show both theoretically and empirically that this method converges quickly to an optimal coupling, essentially at the same computational cost of running vanilla Sinkhorn in each pair of states. Along the way, we point out that our optimal transport distance exactly matches the common notion of bisimulation metrics between Markov chains, and thus our results also apply to computing such metrics, and in fact our algorithm turns out to be significantly more efficient than the best known methods developed so far for this purpose.

cs.LG↗

arXiv2024

Planning with a Learned Policy Basis to Optimally Solve Complex Tasks

Conventional reinforcement learning (RL) methods can successfully solve a wide range of sequential decision problems. However, learning policies that can generalize predictably across multiple tasks in a setting with non-Markovian reward specifications is a challenging problem. We propose to use successor features to learn a policy basis so that each (sub)policy in it solves a well-defined subproblem. In a task described by a finite state automaton (FSA) that involves the same set of subproblems, the combination of these (sub)policies can then be used to generate an optimal solution without additional learning. In contrast to other methods that combine (sub)policies via planning, our method asymptotically attains global optimality, even in stochastic environments.

cs.LG↗

arXiv2023

Asymmetric Norms to Approximate the Minimum Action Distance

This paper presents a state representation for reward-free Markov decision processes. The idea is to learn, in a self-supervised manner, an embedding space where distances between pairs of embedded states correspond to the minimum number of actions needed to transition between them. Unlike previous methods, our approach incorporates an asymmetric norm parametrization, enabling accurate approximations of minimum action distances in environments with inherent asymmetry. We show how this representation can be leveraged to learn goal-conditioned policies, providing a notion of similarity between states and goals and a useful heuristic distance to guide planning. To validate our approach, we conduct empirical experiments on both symmetric and asymmetric environments. Our results show that our asymmetric norm parametrization performs comparably to symmetric norms in symmetric environments and surpasses symmetric norms in asymmetric environments.

cs.LG↗