Hierarchical decomposition of control is unavoidable in large dynamical systems. In reinforcement learning (RL), it is usually solved with subgoals defined at higher policy levels and achieved at lower policy levels. Reaching these goals can take a substantial amount of time, during which it is not verified whether they are still worth pursuing. However, due to the randomness of the environment, these goals may become obsolete. In this paper, we address this gap in the state-of-the-art approaches and propose a method in which the validity of higher-level actions (thus lower-level goals) is constantly verified at the higher level. If the actions, i.e. lower Figure 1: Emergency action termination in hierarchical RL: level goals, become inadequate, they are replaced by more appropriate Left: Pursuing the long-term goal, the agent defines a shortterm ones. This way we combine the advantages of hierarchical one. Middle: When reaching the short-term goal, the RL, which is fast training, and flat RL, which is immediate reactivity.

The electrification of shared mobility has become popular across the globe. Many cities have their new shared e-mobility systems deployed, with continuously expanding coverage from central areas to the city edges. A key challenge in the operation of these systems is fleet rebalancing, i.e., how EVs should be repositioned to better satisfy future demand. This is particularly challenging in the context of expanding systems, because i) the range of the EVs is limited while charging time is typically long, which constrain the viable rebalancing operations; and ii) the EV stations in the system are dynamically changing, i.e., the legitimate targets for rebalancing operations can vary over time. We tackle these challenges by first investigating rich sets of data collected from a real-world shared e-mobility system for one year, analyzing the operation model, usage patterns and expansion dynamics of this new mobility mode. With the learned knowledge we design a high-fidelity simulator, which is able to abstract key operation details of EV sharing at fine granularity. Then we model the rebalancing task for shared e-mobility systems under continuous expansion as a Multi-Agent Reinforcement Learning (MARL) problem, which directly takes the range and charging properties of the EVs into account. We further propose a novel policy optimization approach with action cascading, which is able to cope with the expansion dynamics and solve the formulated MARL. We evaluate the proposed approach extensively, and experimental results show that our approach outperforms the state-of-the-art, offering significant performance gain in both satisfied demand and net revenue. A promising trend of future mobility is electric and shared. For instance, Figure 1 shows the spatial distribution of station occupancy rate (ratio and ride-sharing services [10], [11], and the de facto solution of parked vehicles to the total available space) in a real-world is to rebalance the fleet during operation.

In recent years, advances in deep learning have resulted in a plethora of successes in the use of reinforcement learning (RL) to solve complex sequential decision tasks with high-dimensional inputs. However, existing systems lack the necessary mechanisms to provide humans with a holistic view of their competence, presenting an impediment to their adoption, particularly in critical applications where the decisions an agent makes can have significant consequences. Yet, existing RL-based systems are essentially competency-unaware in that they lack the necessary interpretation mechanisms to allow human operators to have an insightful, holistic view of their competency. In this paper, we extend a recently-proposed framework for explainable RL that is based on analyses of "interestingness." Our new framework provides various measures of RL agent competence stemming from interestingness analysis and is applicable to a wide range of RL algorithms. We also propose novel mechanisms for assessing RL agents' competencies that: 1) identify agent behavior patterns and competency-controlling conditions by clustering agent behavior traces solely using interestingness data; and 2) identify the task elements mostly responsible for an agent's behavior, as measured through interestingness, by performing global and local analyses using SHAP values. Overall, our tools provide insights about RL agent competence, both their capabilities and limitations, enabling users to make more informed decisions about interventions, additional training, and other interactions in collaborative human-machine settings.

Advances in reinforcement learning (RL) often rely on massive compute resources and remain notoriously sample inefficient. In contrast, the human brain is able to efficiently learn effective control strategies using limited resources. This raises the question whether insights from neuroscience can be used to improve current RL methods. Predictive processing is a popular theoretical framework which maintains that the human brain is actively seeking to minimize surprise. We show that recurrent neural networks which predict their own sensory states can be leveraged to minimise surprise, yielding substantial gains in cumulative reward. Specifically, we present the Predictive Processing Proximal Policy Optimization (P4O) agent; an actor-critic reinforcement learning agent that applies predictive processing to a recurrent variant of the PPO algorithm by integrating a world model in its hidden state. P4O significantly outperforms a baseline recurrent variant of the PPO algorithm on multiple Atari games using a single GPU. It also outperforms other state-of-the-art agents given the same wall-clock time and exceeds human gamer performance on multiple games including Seaquest, which is a particularly challenging environment in the Atari domain. Altogether, our work underscores how insights from the field of neuroscience may support the development of more capable and efficient artificial agents.

Inspired by progress in large-scale language modeling, we apply a similar approach towards building a single generalist agent beyond the realm of text outputs. The agent, which we refer to as Gato, works as a multi-modal, multi-task, multi-embodiment generalist policy. The same network with the same weights can play Atari, caption images, chat, stack blocks with a real robot arm and much more, deciding based on its context whether to output text, joint torques, button presses, or other tokens. In this report we describe the model and the data, and document the current capabilities of Gato. What is the capital of France?

Offline goal-conditioned reinforcement learning (GCRL) promises general-purpose skill learning in the form of reaching diverse goals from purely offline datasets. We propose $\textbf{Go}$al-conditioned $f$-$\textbf{A}$dvantage $\textbf{R}$egression (GoFAR), a novel regression-based offline GCRL algorithm derived from a state-occupancy matching perspective; the key intuition is that the goal-reaching task can be formulated as a state-occupancy matching problem between a dynamics-abiding imitator agent and an expert agent that directly teleports to the goal. In contrast to prior approaches, GoFAR does not require any hindsight relabeling and enjoys uninterleaved optimization for its value and policy networks. These distinct features confer GoFAR with much better offline performance and stability as well as statistical performance guarantee that is unattainable for prior methods. Furthermore, we demonstrate that GoFAR's training objectives can be re-purposed to learn an agent-independent goal-conditioned planner from purely offline source-domain data, which enables zero-shot transfer to new target domains. Through extensive experiments, we validate GoFAR's effectiveness in various problem settings and tasks, significantly outperforming prior state-of-art. Notably, on a real robotic dexterous manipulation task, while no other method makes meaningful progress, GoFAR acquires complex manipulation behavior that successfully accomplishes diverse goals.

Recent advancements in the rollout of 5G and 6G have led to the emergence of a new range of latency-critical applications delivered via a Network Function Virtualization (NFV) enabled paradigm of flexible and softwarized communication networks. Evolving verticals like telecommunications, smart grid, virtual reality (VR), industry 4.0, automated vehicles, etc. are driven by the vision of low latency and high reliability, and there is a wide gap to efficiently bridge the Quality of Service (QoS) constraints for both the service providers and the end-user. In this work, we look to tackle the over-provisioning of latency-critical services by proposing a proactive SLA management framework leveraging Graph Neural Networks (GNN) and Deep Reinforcement Learning (DRL) to balance the trade-off between efficiency and reliability. To summarize our key contributions: 1) we compose a graph-based spatio-temporal multivariate time-series forecasting model with multiple time-step predictions in a multi-output scenario, delivering 74.62% improved performance over the established baseline state-of-art model on the use-case; and 2) we leverage realistic SLA definitions for the use-case to achieve a dynamic SLA-aware oversight for scaling policy management with DRL.

Probabilistic Boolean Networks have been proposed for estimating the behaviour of dynamical systems as they combine rule-based modelling with uncertainty principles. Inferring PBNs directly from gene data is challenging however, especially when data is costly to collect and/or noisy, e.g., in the case of gene expression profile data. In this paper, we present a reproducible method for inferring PBNs directly from real gene expression data measurements taken when the system was at a steady state. The steady-state dynamics of PBNs is of special interest in the analysis of biological machinery. The proposed approach does not rely on reconstructing the state evolution of the network, which is computationally intractable for larger networks. We demonstrate the method on samples of real gene expression profiling data from a well-known study on metastatic melanoma. The pipeline is implemented using Python and we make it publicly available.

As an efficient way to integrate multiple distributed energy resources and the user side, a microgrid is mainly faced with the problems of small-scale volatility, uncertainty, intermittency and demand-side uncertainty of DERs. The traditional microgrid has a single form and cannot meet the flexible energy dispatch between the complex demand side and the microgrid. In response to this problem, the overall environment of wind power, thermostatically controlled loads, energy storage systems, price-responsive loads and the main grid is proposed. Secondly, the centralized control of the microgrid operation is convenient for the control of the reactive power and voltage of the distributed power supply and the adjustment of the grid frequency. However, there is a problem in that the flexible loads aggregate and generate peaks during the electricity price valley. The existing research takes into account the power constraints of the microgrid and fails to ensure a sufficient supply of electric energy for a single flexible load. This paper considers the response priority of each unit component of TCLs and ESSs on the basis of the overall environment operation of the microgrid so as to ensure the power supply of the flexible load of the microgrid and save the power input cost to the greatest extent. Finally, the simulation optimization of the environment can be expressed as a Markov decision process process. It combines two stages of offline and online operations in the training process. The addition of multiple threads with the lack of historical data learning leads to low learning efficiency. The asynchronous advantage actor-critic with the experience replay pool memory library is added to solve the data correlation and nonstatic distribution problems during training.

We investigate the use of animal videos (observations) to improve Reinforcement Learning (RL) efficiency and performance in navigation tasks with sparse rewards. Motivated by theoretical considerations, we make use of weighted policy optimization for off-policy RL and describe the main challenges when learning from animal videos. We propose solutions and test our ideas on a series of 2D navigation tasks. We show how our methods can leverage animal videos to improve performance over RL algorithms that do not leverage such observations.