Tensor graph superoptimisation systems perform a sequence of subgraph substitution to neural networks, to find the optimal computation graph structure. Such a graph transformation process naturally falls into the framework of sequential decision-making, and existing systems typically employ a greedy search approach, which cannot explore the whole search space as it cannot tolerate a temporary loss of performance. In this paper, we address the tensor graph superoptimisation problem by exploring an alternative search approach, reinforcement learning (RL). Our proposed approach, X-RLflow, can learn to perform neural network dataflow graph rewriting, which substitutes a subgraph one at a time. X-RLflow is based on a model-free RL agent that uses a graph neural network (GNN) to encode the target computation graph and outputs a transformed computation graph iteratively. We show that our approach can outperform state-of-the-art superoptimisation systems over a range of deep learning models and achieve by up to 40% on those that are based on transformer-style architectures.

Aiming at selecting a small subset of nodes with maximum influence on networks, the Influence Maximization (IM) problem has been extensively studied. Since it is #P-hard to compute the influence spread given a seed set, the state-of-the-art methods, including heuristic and approximation algorithms, faced with great difficulties such as theoretical guarantee, time efficiency, generalization, etc. This makes it unable to adapt to large-scale networks and more complex applications. On the other side, with the latest achievements of Deep Reinforcement Learning (DRL) in artificial intelligence and other fields, lots of works have been focused on exploiting DRL to solve combinatorial optimization problems. Inspired by this, we propose a novel end-to-end DRL framework, ToupleGDD, to address the IM problem in this paper, which incorporates three coupled graph neural networks for network embedding and double deep Q-networks for parameters learning. Previous efforts to solve IM problem with DRL trained their models on subgraphs of the whole network, and then tested on the whole graph, which makes the performance of their models unstable among different networks. However, our model is trained on several small randomly generated graphs with a small budget, and tested on completely different networks under various large budgets, which can obtain results very close to IMM and better results than OPIM-C on several datasets, and shows strong generalization ability. Finally, we conduct a large number of experiments on synthetic and realistic datasets, and experimental results prove the effectiveness and superiority of our model.

Autonomous navigation in crowded spaces poses a challenge for mobile robots due to the highly dynamic, partially observable environment. Occlusions are highly prevalent in such settings due to a limited sensor field of view and obstructing human agents. Previous work has shown that observed interactive behaviors of human agents can be used to estimate potential obstacles despite occlusions. We propose integrating such social inference techniques into the planning pipeline. We use a variational autoencoder with a specially designed loss function to learn representations that are meaningful for occlusion inference. This work adopts a deep reinforcement learning approach to incorporate the learned representation for occlusion-aware planning. In simulation, our occlusion-aware policy achieves comparable collision avoidance performance to fully observable navigation by estimating agents in occluded spaces. We demonstrate successful policy transfer from simulation to the real-world Turtlebot 2i. To the best of our knowledge, this work is the first to use social occlusion inference for crowd navigation.

Next-generation Wi-Fi networks are looking forward to introducing new features like multi-link operation (MLO) to both achieve higher throughput and lower latency. However, given the limited number of available channels, the use of multiple links by a group of contending Basic Service Sets (BSSs) can result in higher interference and channel contention, thus potentially leading to lower performance and reliability. In such a situation, it could be better for all contending BSSs to use less links if that contributes to reduce channel access contention. Recently, reinforcement learning (RL) has proven its potential for optimizing resource allocation in wireless networks. However, the independent operation of each wireless network makes difficult -- if not almost impossible -- for each individual network to learn a good configuration. To solve this issue, in this paper, we propose the use of a Federated Reinforcement Learning (FRL) framework, i.e., a collaborative machine learning approach to train models across multiple distributed agents without exchanging data, to collaboratively learn the the best MLO-Link Allocation (LA) strategy by a group of neighboring BSSs. The simulation results show that the FRL-based decentralized MLO-LA strategy achieves a better throughput fairness, and so a higher reliability -- because it allows the different BSSs to find a link allocation strategy which maximizes the minimum achieved data rate -- compared to fixed, random and RL-based MLO-LA schemes.

Nowadays, the application of microgrids (MG) with renewable energy is becoming more and more extensive, which creates a strong need for dynamic energy management. In this paper, deep reinforcement learning (DRL) is applied to learn an optimal policy for making joint energy dispatch (ED) and unit commitment (UC) decisions in an isolated MG, with the aim for reducing the total power generation cost on the premise of ensuring the supply-demand balance. In order to overcome the challenge of discrete-continuous hybrid action space due to joint ED and UC, we propose a DRL algorithm, i.e., the hybrid action finite-horizon DDPG (HAFH-DDPG), that seamlessly integrates two classical DRL algorithms, i.e., deep Q-network (DQN) and deep deterministic policy gradient (DDPG), based on a finite-horizon dynamic programming (DP) framework. Moreover, a diesel generator (DG) selection strategy is presented to support a simplified action space for reducing the computation complexity of this algorithm. Finally, the effectiveness of our proposed algorithm is verified through comparison with several baseline algorithms by experiments with real-world data set.

Centralized training with decentralized execution (CTDE) is a widely-used learning paradigm that has achieved significant success in complex tasks. However, partial observability issues and the absence of effectively shared signals between agents often limit its effectiveness in fostering cooperation. While communication can address this challenge, it simultaneously reduces the algorithm's practicality. Drawing inspiration from human team cooperative learning, we propose a novel paradigm that facilitates a gradual shift from explicit communication to tacit cooperation. In the initial training stage, we promote cooperation by sharing relevant information among agents and concurrently reconstructing this information using each agent's local trajectory. We then combine the explicitly communicated information with the reconstructed information to obtain mixed information. Throughout the training process, we progressively reduce the proportion of explicitly communicated information, facilitating a seamless transition to fully decentralized execution without communication. Experimental results in various scenarios demonstrate that the performance of our method without communication can approaches or even surpasses that of QMIX and communication-based methods.

In this paper, we investigate the scheduling issue of diesel generators (DGs) in an Internet of Things (IoT)-Driven isolated microgrid (MG) by deep reinforcement learning (DRL). The renewable energy is fully exploited under the uncertainty of renewable generation and load demand. The DRL agent learns an optimal policy from history renewable and load data of previous days, where the policy can generate real-time decisions based on observations of past renewable and load data of previous hours collected by connected sensors. The goal is to reduce operating cost on the premise of ensuring supply-demand balance. In specific, a novel finite-horizon partial observable Markov decision process (POMDP) model is conceived considering the spinning reserve. In order to overcome the challenge of discrete-continuous hybrid action space due to the binary DG switching decision and continuous energy dispatch (ED) decision, a DRL algorithm, namely the hybrid action finite-horizon RDPG (HAFH-RDPG), is proposed. HAFH-RDPG seamlessly integrates two classical DRL algorithms, i.e., deep Q-network (DQN) and recurrent deterministic policy gradient (RDPG), based on a finite-horizon dynamic programming (DP) framework. Extensive experiments are performed with real-world data in an IoT-driven MG to evaluate the capability of the proposed algorithm in handling the uncertainty due to inter-hour and inter-day power fluctuation and to compare its performance with those of the benchmark algorithms. J. Qi, L. Lei, and S. X. Yang are with the School of Engineering, University of Guelph, Guelph, ON N1G 2W1, Canada (e-mail: jiaju@uoguelph.ca; K. Zheng is with the College of Electrical Engineering and Computer Sciences, Ningbo University, Ningbo, 315211, China.

Model Predictive Control (MPC) is attracting tremendous attention in the autonomous driving task as a powerful control technique. The success of an MPC controller strongly depends on an accurate internal dynamics model. However, the static parameters, usually learned by system identification, often fail to adapt to both internal and external perturbations in real-world scenarios. In this paper, we firstly (1) reformulate the problem as a Partially Observed Markov Decision Process (POMDP) that absorbs the uncertainties into observations and maintains Markov property into hidden states; and (2) learn a recurrent policy continually adapting the parameters of the dynamics model via Recurrent Reinforcement Learning (RRL) for optimal and adaptive control; and (3) finally evaluate the proposed algorithm (referred as $\textit{MPC-RRL}$) in CARLA simulator and leading to robust behaviours under a wide range of perturbations.

Many decision-making problems feature multiple objectives. In such problems, it is not always possible to know the preferences of a decision-maker for different objectives. However, it is often possible to observe the behavior of decision-makers. In multi-objective decision-making, preference inference is the process of inferring the preferences of a decision-maker for different objectives. This research proposes a Dynamic Weight-based Preference Inference (DWPI) algorithm that can infer the preferences of agents acting in multi-objective decision-making problems, based on observed behavior trajectories in the environment. The proposed method is evaluated on three multi-objective Markov decision processes: Deep Sea Treasure, Traffic, and Item Gathering. The performance of the proposed DWPI approach is compared to two existing preference inference methods from the literature, and empirical results demonstrate significant improvements compared to the baseline algorithms, in terms of both time requirements and accuracy of the inferred preferences. The Dynamic Weight-based Preference Inference algorithm also maintains its performance when inferring preferences for sub-optimal behavior demonstrations. In addition to its impressive performance, the Dynamic Weight-based Preference Inference algorithm does not require any interactions during training with the agent whose preferences are inferred, all that is required is a trajectory of observed behavior.

In this paper, we introduce Latent Go-Explore (LGE), a simple and general approach based on the Go-Explore paradigm for exploration in reinforcement learning (RL). Go-Explore was initially introduced with a strong domain knowledge constraint for partitioning the state space into cells. However, in most real-world scenarios, drawing domain knowledge from raw observations is complex and tedious. If the cell partitioning is not informative enough, Go-Explore can completely fail to explore the environment. We argue that the Go-Explore approach can be generalized to any environment without domain knowledge and without cells by exploiting a learned latent representation. Thus, we show that LGE can be flexibly combined with any strategy for learning a latent representation. Our results indicate that LGE, although simpler than Go-Explore, is more robust and outperforms state-of-the-art algorithms in terms of pure exploration on multiple hard-exploration environments including Montezuma's Revenge. The LGE implementation is available as open-source at https://github.com/qgallouedec/lge.