Denis Steckelmacher

Qingshuang Sun, Denis Steckelmacher, Yuan Yao et al.

Communication plays a vital role in multi-agent systems, fostering collaboration and coordination. However, in real-world scenarios where communication is bandwidth-limited, existing multi-agent reinforcement learning (MARL) algorithms often provide agents with a binary choice: either transmitting a fixed number of bytes or no information at all. This limitation hinders the ability to effectively utilize the available bandwidth. To overcome this challenge, we present the Dynamic Size Message Scheduling (DSMS) method, which introduces a finer-grained approach to scheduling by considering the actual size of the information to be exchanged. Our contribution lies in adaptively adjusting message sizes using Fourier transform-based compression techniques, enabling agents to tailor their messages to match the allocated bandwidth while striking a balance between information loss and transmission efficiency. Receiving agents can reliably decompress the messages using the inverse Fourier transform. Experimental results demonstrate that DSMS significantly improves performance in multi-agent cooperative tasks by optimizing the utilization of bandwidth and effectively balancing information value.

Hélène Plisnier, Denis Steckelmacher, Jeroen Willems et al.

Many instances of similar or almost-identical industrial machines or tools are often deployed at once, or in quick succession. For instance, a particular model of air compressor may be installed at hundreds of customers. Because these tools perform distinct but highly similar tasks, it is interesting to be able to quickly produce a high-quality controller for machine $N+1$ given the controllers already produced for machines $1..N$. This is even more important when the controllers are learned through Reinforcement Learning, as training takes time, energy and other resources. In this paper, we apply Policy Intersection, a Policy Shaping method, to help a Reinforcement Learning agent learn to solve a new variant of a compressors control problem faster, by transferring knowledge from several previously learned controllers. We show that our approach outperforms loading an old controller, and significantly improves performance in the long run.

Senne Deproost, Denis Steckelmacher, Ann Nowé

Despite many successful attempts at explaining Deep Reinforcement Learning policies using distillation, it remains difficult to balance the performance-interpretability trade-off and select a fitting surrogate model. In addition to this, traditional distillation only minimizes the distance between the behavior of the original and the surrogate policy while other RL-specific components such as action value are disregarded. To solve this, we introduce a new model-agnostic method called Critic-Driven Voronoi State Partitioning, which partitions a black box control policy into regions where a simple class of model can be optimized using gradient descent. By exploiting the critic value network of the original policy, we iteratively introduce new subpolicies in regions with insufficient value, standing in for a measure of policy complexity. The partitioning, a Voronoi quantizer, uses nearest neighbor lookups to assign a linear function to each point in the state space resulting in a cell-like diagram. We validate our approach on several well known benchmarks and proof that this distillation approaches the original policy using a reasonable sized set of linear functions.

Denis Steckelmacher, Hélène Plisnier, Diederik M. Roijers et al.

Value-based reinforcement-learning algorithms provide state-of-the-art results in model-free discrete-action settings, and tend to outperform actor-critic algorithms. We argue that actor-critic algorithms are limited by their need for an on-policy critic. We propose Bootstrapped Dual Policy Iteration (BDPI), a novel model-free reinforcement-learning algorithm for continuous states and discrete actions, with an actor and several off-policy critics. Off-policy critics are compatible with experience replay, ensuring high sample-efficiency, without the need for off-policy corrections. The actor, by slowly imitating the average greedy policy of the critics, leads to high-quality and state-specific exploration, which we compare to Thompson sampling. Because the actor and critics are fully decoupled, BDPI is remarkably stable, and unusually robust to its hyper-parameters. BDPI is significantly more sample-efficient than Bootstrapped DQN, PPO, and ACKTR, on discrete, continuous and pixel-based tasks. Source code: https://github.com/vub-ai-lab/bdpi.

Senne Deproost, Denis Steckelmacher, Ann Nowé

With Deep Reinforcement Learning (DRL) being increasingly considered for the control of real-world systems, the lack of transparency of the neural network at the core of RL becomes a concern. Programmatic Reinforcement Learning (PRL) is able to to create representations of this black-box in the form of source code, not only increasing the explainability of the controller but also allowing for user adaptations. However, these methods focus on distilling a black-box policy into a program and do so after learning using the Mean Squared Error between produced and wanted behaviour, discarding other elements of the RL algorithm. The distilled policy may therefore perform significantly worse than the black-box learned policy. In this paper, we propose to directly learn a program as the policy of an RL agent. We build on TD3 and use its critics as the basis of the objective function of a genetic algorithm that syntheses the program. Our approach builds the program during training, as opposed to after the fact. This steers the program to actual high rewards, instead of a simple Mean Squared Error. Also, our approach leverages the TD3 critics to achieve high sample-efficiency, as opposed to pure genetic methods that rely on Monte-Carlo evaluations. Our experiments demonstrate the validity, explainability and sample-efficiency of our approach in a simple gridworld environment.

Youri Coppens, Denis Steckelmacher, Catholijn M. Jonker et al.

Today's advanced Reinforcement Learning algorithms produce black-box policies, that are often difficult to interpret and trust for a person. We introduce a policy distilling algorithm, building on the CN2 rule mining algorithm, that distills the policy into a rule-based decision system. At the core of our approach is the fact that an RL process does not just learn a policy, a mapping from states to actions, but also produces extra meta-information, such as action values indicating the quality of alternative actions. This meta-information can indicate whether more than one action is near-optimal for a certain state. We extend CN2 to make it able to leverage knowledge about equally-good actions to distill the policy into fewer rules, increasing its interpretability by a person. Then, to ensure that the rules explain a valid, non-degenerate policy, we introduce a refinement algorithm that fine-tunes the rules to obtain good performance when executed in the environment. We demonstrate the applicability of our algorithm on the Mario AI benchmark, a complex task that requires modern reinforcement learning algorithms including neural networks. The explanations we produce capture the learned policy in only a few rules, that allow a person to understand what the black-box agent learned. Source code: https://gitlab.ai.vub.ac.be/yocoppen/svcn2

Hélène Plisnier, Denis Steckelmacher, Diederik Roijers et al.

For a robot to learn a good policy, it often requires expensive equipment (such as sophisticated sensors) and a prepared training environment conducive to learning. However, it is seldom possible to perfectly equip robots for economic reasons, nor to guarantee ideal learning conditions, when deployed in real-life environments. A solution would be to prepare the robot in the lab environment, when all necessary material is available to learn a good policy. After training in the lab, the robot should be able to get by without the expensive equipment that used to be available to it, and yet still be guaranteed to perform well on the field. The transition between the lab (source) and the real-world environment (target) is related to transfer learning, where the state-space between the source and target tasks differ. We tackle a simulated task with continuous states and discrete actions presenting this challenge, using Bootstrapped Dual Policy Iteration, a model-free actor-critic reinforcement learning algorithm, and Policy Shaping. Specifically, we train a BDPI agent, embodied by a virtual robot performing a task in the V-Rep simulator, sensing its environment through several proximity sensors. The resulting policy is then used by a second agent learning the same task in the same environment, but with camera images as input. The goal is to obtain a policy able to perform the task relying on merely camera images.

Hélène Plisnier, Denis Steckelmacher, Diederik M. Roijers et al.

Actor-critic algorithms learn an explicit policy (actor), and an accompanying value function (critic). The actor performs actions in the environment, while the critic evaluates the actor's current policy. However, despite their stability and promising convergence properties, current actor-critic algorithms do not outperform critic-only ones in practice. We believe that the fact that the critic learns Q^pi, instead of the optimal Q-function Q*, prevents state-of-the-art robust and sample-efficient off-policy learning algorithms from being used. In this paper, we propose an elegant solution, the Actor-Advisor architecture, in which a Policy Gradient actor learns from unbiased Monte-Carlo returns, while being shaped (or advised) by the Softmax policy arising from an off-policy critic. The critic can be learned independently from the actor, using any state-of-the-art algorithm. Being advised by a high-quality critic, the actor quickly and robustly learns the task, while its use of the Monte-Carlo return helps overcome any bias the critic may have. In addition to a new Actor-Critic formulation, the Actor-Advisor, a method that allows an external advisory policy to shape a Policy Gradient actor, can be applied to many other domains. By varying the source of advice, we demonstrate the wide applicability of the Actor-Advisor to three other important subfields of RL: safe RL with backup policies, efficient leverage of domain knowledge, and transfer learning in RL. Our experimental results demonstrate the benefits of the Actor-Advisor compared to state-of-the-art actor-critic methods, illustrate its applicability to the three other application scenarios listed above, and show that many important challenges of RL can now be solved using a single elegant solution.

Axel Abels, Diederik M. Roijers, Tom Lenaerts et al.

Many real-world decision problems are characterized by multiple conflicting objectives which must be balanced based on their relative importance. In the dynamic weights setting the relative importance changes over time and specialized algorithms that deal with such change, such as a tabular Reinforcement Learning (RL) algorithm by Natarajan and Tadepalli (2005), are required. However, this earlier work is not feasible for RL settings that necessitate the use of function approximators. We generalize across weight changes and high-dimensional inputs by proposing a multi-objective Q-network whose outputs are conditioned on the relative importance of objectives and we introduce Diverse Experience Replay (DER) to counter the inherent non-stationarity of the Dynamic Weights setting. We perform an extensive experimental evaluation and compare our methods to adapted algorithms from Deep Multi-Task/Multi-Objective Reinforcement Learning and show that our proposed network in combination with DER dominates these adapted algorithms across weight change scenarios and problem domains.

Hélène Plisnier, Denis Steckelmacher, Tim Brys et al.

Many currently deployed Reinforcement Learning agents work in an environment shared with humans, be them co-workers, users or clients. It is desirable that these agents adjust to people's preferences, learn faster thanks to their help, and act safely around them. We argue that most current approaches that learn from human feedback are unsafe: rewarding or punishing the agent a-posteriori cannot immediately prevent it from wrong-doing. In this paper, we extend Policy Gradient to make it robust to external directives, that would otherwise break the fundamentally on-policy nature of Policy Gradient. Our technique, Directed Policy Gradient (DPG), allows a teacher or backup policy to override the agent before it acts undesirably, while allowing the agent to leverage human advice or directives to learn faster. Our experiments demonstrate that DPG makes the agent learn much faster than reward-based approaches, while requiring an order of magnitude less advice.

Denis Steckelmacher, Diederik M. Roijers, Anna Harutyunyan et al.

Many real-world reinforcement learning problems have a hierarchical nature, and often exhibit some degree of partial observability. While hierarchy and partial observability are usually tackled separately (for instance by combining recurrent neural networks and options), we show that addressing both problems simultaneously is simpler and more efficient in many cases. More specifically, we make the initiation set of options conditional on the previously-executed option, and show that options with such Option-Observation Initiation Sets (OOIs) are at least as expressive as Finite State Controllers (FSCs), a state-of-the-art approach for learning in POMDPs. OOIs are easy to design based on an intuitive description of the task, lead to explainable policies and keep the top-level and option policies memoryless. Our experiments show that OOIs allow agents to learn optimal policies in challenging POMDPs, while being much more sample-efficient than a recurrent neural network over options.

Denis Steckelmacher, Peter Vrancx

This paper explores the performance of fitted neural Q iteration for reinforcement learning in several partially observable environments, using three recurrent neural network architectures: Long Short-Term Memory, Gated Recurrent Unit and MUT1, a recurrent neural architecture evolved from a pool of several thousands candidate architectures. A variant of fitted Q iteration, based on Advantage values instead of Q values, is also explored. The results show that GRU performs significantly better than LSTM and MUT1 for most of the problems considered, requiring less training episodes and less CPU time before learning a very good policy. Advantage learning also tends to produce better results.