This article is part of Demystifying AI, a series of posts that (try to) disambiguate the jargon and myths surrounding AI. Deep reinforcement learning is one of the most interesting branches of artificial intelligence. It is behind some of the most remarkable achievements of the AI community, including beating human champions at board and video games, self-driving cars, robotics, and AI hardware design. Deep reinforcement learning leverages the learning capacity of deep neural networks to tackle problems that were too complex for classic RL techniques. Deep reinforcement learning is much more complicated than the other branches of machine learning.

Many off-policy prediction learning algorithms have been proposed in the past decade, but it remains unclear which algorithms learn faster than others. We empirically compare 11 off-policy prediction learning algorithms with linear function approximation on two small tasks: the Rooms task, and the High Variance Rooms task. The tasks are designed such that learning fast in them is challenging. In the Rooms task, the product of importance sampling ratios can be as large as $2^{14}$ and can sometimes be two. To control the high variance caused by the product of the importance sampling ratios, step size should be set small, which in turn slows down learning. The High Variance Rooms task is more extreme in that the product of the ratios can become as large as $2^{14}\times 25$. This paper builds upon the empirical study of off-policy prediction learning algorithms by Ghiassian and Sutton (2021). We consider the same set of algorithms as theirs and employ the same experimental methodology. The algorithms considered are: Off-policy TD($\lambda$), five Gradient-TD algorithms, two Emphatic-TD algorithms, Tree Backup($\lambda$), Vtrace($\lambda$), and ABTD($\zeta$). We found that the algorithms' performance is highly affected by the variance induced by the importance sampling ratios. The data shows that Tree Backup($\lambda$), Vtrace($\lambda$), and ABTD($\zeta$) are not affected by the high variance as much as other algorithms but they restrict the effective bootstrapping parameter in a way that is too limiting for tasks where high variance is not present. We observed that Emphatic TD($\lambda$) tends to have lower asymptotic error than other algorithms, but might learn more slowly in some cases. We suggest algorithms for practitioners based on their problem of interest, and suggest approaches that can be applied to specific algorithms that might result in substantially improved algorithms.

A lowering in the cost of batteries and solar PV systems has led to a high uptake of solar battery home systems. In this work, we use the deep deterministic policy gradient algorithm to optimise the charging and discharging behaviour of a battery within such a system. Our approach outputs a continuous action space when it charges and discharges the battery, and can function well in a stochastic environment. We show good performance of this algorithm by lowering the expenditure of a single household on electricity to almost \$1AUD for large batteries across selected weeks within a year.

Learning to solve sparse-reward reinforcement learning problems is difficult, due to the lack of guidance towards the goal. But in some problems, prior knowledge can be used to augment the learning process. Reward shaping is a way to incorporate prior knowledge into the original reward function in order to speed up the learning. While previous work has investigated the use of expert knowledge to generate potential functions, in this work, we study whether we can use a search algorithm(A*) to automatically generate a potential function for reward shaping in Sokoban, a well-known planning task. The results showed that learning with shaped reward function is faster than learning from scratch. Our results indicate that distance functions could be a suitable function for Sokoban. This work demonstrates the possibility of solving multiple instances with the help of reward shaping. The result can be compressed into a single policy, which can be seen as the first phrase towards training a general policy that is able to solve unseen instances.

We present in this work an approach to reduce the communication of information needed on a multi-agent learning system inspired by Event Triggered Control (ETC) techniques. We consider a baseline scenario of a distributed Q-learning problem on a Markov Decision Process (MDP). Following an event-based approach, N agents explore the MDP and communicate experiences to a central learner only when necessary, which performs updates of the actor Q functions. We analyse the convergence guarantees retained with respect to a regular Q-learning algorithm, and present experimental results showing that event-based communication results in a substantial reduction of data transmission rates in such distributed systems. Additionally, we discuss what effects (desired and undesired) these event-based approaches have on the learning processes studied, and how they can be applied to more complex multi-agent learning systems.

Inverse Reinforcement Learning (IRL) is attractive in scenarios where reward engineering can be tedious. However, prior IRL algorithms use on-policy transitions, which require intensive sampling from the current policy for stable and optimal performance. This limits IRL applications in the real world, where environment interactions can become highly expensive. To tackle this problem, we present Off-Policy Inverse Reinforcement Learning (OPIRL), which (1) adopts off-policy data distribution instead of on-policy and enables significant reduction of the number of interactions with the environment, (2) learns a stationary reward function that is transferable with high generalization capabilities on changing dynamics, and (3) leverages mode-covering behavior for faster convergence. We demonstrate that our method is considerably more sample efficient and generalizes to novel environments through the experiments. Our method achieves better or comparable results on policy performance baselines with significantly fewer interactions. Furthermore, we empirically show that the recovered reward function generalizes to different tasks where prior arts are prone to fail.

Despite the potential of active inference for visual-based control, learning the model and the preferences (priors) while interacting with the environment is challenging. Here, we study the performance of a deep active inference (dAIF) agent on OpenAI's car racing benchmark, where there is no access to the car's state. The agent learns to encode the world's state from high-dimensional input through unsupervised representation learning. State inference and control are learned end-to-end by optimizing the expected free energy. Results show that our model achieves comparable performance to deep Q-learning. However, vanilla dAIF does not reach state-of-the-art performance compared to other world model approaches. Hence, we discuss the current model implementation's limitations and potential architectures to overcome them.

In a ride-hailing system, an optimal relocation of vacant vehicles can significantly reduce fleet idling time and balance the supply-demand distribution, enhancing system efficiency and promoting driver satisfaction and retention. Model-free deep reinforcement learning (DRL) has been shown to dynamically learn the relocating policy by actively interacting with the intrinsic dynamics in large-scale ride-hailing systems. However, the issues of sparse reward signals and unbalanced demand and supply distribution place critical barriers in developing effective DRL models. Conventional exploration strategy (e.g., the $\epsilon$-greedy) may barely work under such an environment because of dithering in low-demand regions distant from high-revenue regions. This study proposes the deep relocating option policy (DROP) that supervises vehicle agents to escape from oversupply areas and effectively relocate to potentially underserved areas. We propose to learn the Laplacian embedding of a time-expanded relocation graph, as an approximation representation of the system relocation policy. The embedding generates task-agnostic signals, which in combination with task-dependent signals, constitute the pseudo-reward function for generating DROPs. We present a hierarchical learning framework that trains a high-level relocation policy and a set of low-level DROPs. The effectiveness of our approach is demonstrated using a custom-built high-fidelity simulator with real-world trip record data. We report that DROP significantly improves baseline models with 15.7% more hourly revenue and can effectively resolve the dithering issue in low-demand areas.

Temporal knowledge graph (TKG) reasoning is a crucial task that has gained increasing research interest in recent years. Most existing methods focus on reasoning at past timestamps to complete the missing facts, and there are only a few works of reasoning on known TKGs to forecast future facts. Compared with the completion task, the forecasting task is more difficult that faces two main challenges: (1) how to effectively model the time information to handle future timestamps? (2) how to make inductive inference to handle previously unseen entities that emerge over time? To address these challenges, we propose the first reinforcement learning method for forecasting. Specifically, the agent travels on historical knowledge graph snapshots to search for the answer. Our method defines a relative time encoding function to capture the timespan information, and we design a novel time-shaped reward based on Dirichlet distribution to guide the model learning. Furthermore, we propose a novel representation method for unseen entities to improve the inductive inference ability of the model. We evaluate our method for this link prediction task at future timestamps. Extensive experiments on four benchmark datasets demonstrate substantial performance improvement meanwhile with higher explainability, less calculation, and fewer parameters when compared with existing state-of-the-art methods.