In this paper we present a deep graph reinforcement learning model to predict and improve the user experience during a live video streaming event, orchestrated by an agent/tracker. We first formulate the user experience prediction problem as a classification task, accounting for the fact that most of the viewers at the beginning of an event have poor quality of experience due to lowbandwidth connections and limited interactions with the tracker. In our model we consider different factors that influence the quality of user experience and train the proposed model on diverse state-action transitions when viewers interact with the tracker. In addition, provided that past events have various user experience characteristics we follow a gradient boosting strategy to compute a global model that learns from different events. Our experiments with three real-world datasets of live video streaming events demonstrate the superiority of the proposed model against several baseline strategies. Moreover, as the majority of the viewers at the beginning of an event has poor experience, we show that our model Figure 1: In a live video streaming event, a viewer periodically can significantly increase the number of viewers with high quality reports the connection bandwidth as well as her quality experience by at least 75% over the first streaming minutes. Our of experience to the tracker which then probes the viewers evaluation datasets and implementation are publicly available at to adjust their connections accordingly.

Learning continuous control in high-dimensional sparse reward settings, such as robotic manipulation, is a challenging problem due to the number of samples often required to obtain accurate optimal value and policy estimates. While many deep reinforcement learning methods have aimed at improving sample efficiency through replay or improved exploration techniques, state of the art actor-critic and policy gradient methods still suffer from the hard exploration problem in sparse reward settings. Motivated by recent successes of value-based methods for approximating state-action values, like RBF-DQN, we explore the potential of value-based reinforcement learning for learning continuous robotic manipulation tasks in multi-task sparse reward settings. On robotic manipulation tasks, we empirically show RBF-DQN converges faster than current state of the art algorithms such as TD3, SAC, and PPO. We also perform ablation studies with RBF-DQN and have shown that some enhancement techniques for vanilla Deep Q learning such as Hindsight Experience Replay (HER) and Prioritized Experience Replay (PER) can also be applied to RBF-DQN. Our experimental analysis suggests that value-based approaches may be more sensitive to data augmentation and replay buffer sample techniques than policy-gradient methods, and that the benefits of these methods for robot manipulation are heavily dependent on the transition dynamics of generated subgoal states.

Although learning from data is effective and has achieved significant milestones, it has many challenges and limitations. Learning from data starts from observations and then proceeds to broader generalizations. This framework is controversial in science, yet it has achieved remarkable engineering successes. This paper reflects on some epistemological issues and some of the limitations of the knowledge discovered in data. The document discusses the common perception that getting more data is the key to achieving better machine learning models from theoretical and practical perspectives. The paper sheds some light on the shortcomings of using generic mathematical theories to describe the process. It further highlights the need for theories specialized in learning from data. While more data leverages the performance of machine learning models in general, the relation in practice is shown to be logarithmic at its best; After a specific limit, more data stabilize or degrade the machine learning models. Recent work in reinforcement learning showed that the trend is shifting away from data-oriented approaches and relying more on algorithms. The paper concludes that learning from data is hindered by many limitations. Hence an approach that has an intensional orientation is needed.

Packet routing is a fundamental problem in communication networks that decides how the packets are directed from their source nodes to their destination nodes through some intermediate nodes. With the increasing complexity of network topology and highly dynamic traffic demand, conventional model-based and rule-based routing schemes show significant limitations, due to the simplified and unrealistic model assumptions, and lack of flexibility and adaption. Adding intelligence to the network control is becoming a trend and the key to achieving high-efficiency network operation. In this paper, we develop a model-free and data-driven routing strategy by leveraging reinforcement learning (RL), where routers interact with the network and learn from the experience to make some good routing configurations for the future. Considering the graph nature of the network topology, we design a multi-agent RL framework in combination with Graph Neural Network (GNN), tailored to the routing problem. Three deployment paradigms, centralized, federated, and cooperated learning, are explored respectively. Simulation results demonstrate that our algorithm outperforms some existing benchmark algorithms in terms of packet transmission delay and affordable load.

Recently introduced distributed zeroth-order optimization (ZOO) algorithms have shown their utility in distributed reinforcement learning (RL). Unfortunately, in the gradient estimation process, almost all of them require random samples with the same dimension as the global variable and/or require evaluation of the global cost function, which may induce high estimation variance for large-scale networks. In this paper, we propose a novel distributed zeroth-order algorithm by leveraging the network structure inherent in the optimization objective, which allows each agent to estimate its local gradient by local cost evaluation independently, without use of any consensus protocol. The proposed algorithm exhibits an asynchronous update scheme, and is designed for stochastic non-convex optimization with a possibly non-convex feasible domain based on the block coordinate descent method. The algorithm is later employed as a distributed model-free RL algorithm for distributed linear quadratic regulator design, where a learning graph is designed to describe the required interaction relationship among agents in distributed learning. We provide an empirical validation of the proposed algorithm to benchmark its performance on convergence rate and variance against a centralized ZOO algorithm.

Reinforcement learning (RL) promises to enable autonomous acquisition of complex behaviors for diverse agents. However, the success of current reinforcement learning algorithms is predicated on an often under-emphasised requirement -- each trial needs to start from a fixed initial state distribution. Unfortunately, resetting the environment to its initial state after each trial requires substantial amount of human supervision and extensive instrumentation of the environment which defeats the purpose of autonomous reinforcement learning. In this work, we propose Value-accelerated Persistent Reinforcement Learning (VaPRL), which generates a curriculum of initial states such that the agent can bootstrap on the success of easier tasks to efficiently learn harder tasks. The agent also learns to reach the initial states proposed by the curriculum, minimizing the reliance on human interventions into the learning. We observe that VaPRL reduces the interventions required by three orders of magnitude compared to episodic RL while outperforming prior state-of-the art methods for reset-free RL both in terms of sample efficiency and asymptotic performance on a variety of simulated robotics problems.

Recent advances in the areas of multimodal machine learning and artificial intelligence (AI) have led to the development of challenging tasks at the intersection of Computer Vision, Natural Language Processing, and Embodied AI. Whereas many approaches and previous survey pursuits have characterised one or two of these dimensions, there has not been a holistic analysis at the center of all three. Moreover, even when combinations of these topics are considered, more focus is placed on describing, e.g., current architectural methods, as opposed to also illustrating high-level challenges and opportunities for the field. In this survey paper, we discuss Embodied Vision-Language Planning (EVLP) tasks, a family of prominent embodied navigation and manipulation problems that jointly use computer vision and natural language. We propose a taxonomy to unify these tasks and provide an in-depth analysis and comparison of the new and current algorithmic approaches, metrics, simulated environments, as well as the datasets used for EVLP tasks. Finally, we present the core challenges that we believe new EVLP works should seek to address, and we advocate for task construction that enables model generalizability and furthers real-world deployment.

Validating the safety of autonomous systems generally requires the use of high-fidelity simulators that adequately capture the variability of real-world scenarios. However, it is generally not feasible to exhaustively search the space of simulation scenarios for failures. Adaptive stress testing (AST) is a method that uses reinforcement learning to find the most likely failure of a system. AST with a deep reinforcement learning solver has been shown to be effective in finding failures across a range of different systems. This approach generally involves running many simulations, which can be very expensive when using a high-fidelity simulator. To improve efficiency, we present a method that first finds failures in a low-fidelity simulator. It then uses the backward algorithm, which trains a deep neural network policy using a single expert demonstration, to adapt the low-fidelity failures to high-fidelity. We have created a series of autonomous vehicle validation case studies that represent some of the ways low-fidelity and high-fidelity simulators can differ, such as time discretization. We demonstrate in a variety of case studies that this new AST approach is able to find failures with significantly fewer high-fidelity simulation steps than are needed when just running AST directly in high-fidelity. As a proof of concept, we also demonstrate AST on NVIDIA's DriveSim simulator, an industry state-of-the-art high-fidelity simulator for finding failures in autonomous vehicles.

We explore the reinforcement learning approach to designing controllers by extensively discussing the case of a quadcopter attitude controller. We provide all details allowing to reproduce our approach, starting with a model of the dynamics of a crazyflie 2.0 under various nominal and non-nominal conditions, including partial motor failures and wind gusts. We develop a robust form of a signal temporal logic to quantitatively evaluate the vehicle's behavior and measure the performance of controllers. The paper thoroughly describes the choices in training algorithms, neural net architecture, hyperparameters, observation space in view of the different performance metrics we have introduced. We discuss the robustness of the obtained controllers, both to partial loss of power for one rotor and to wind gusts and finish by drawing conclusions on practical controller design by reinforcement learning.

This paper presents a novel approach to automated playtesting for the prediction of human player behavior and experience. It has previously been demonstrated that Deep Reinforcement Learning (DRL) game-playing agents can predict both game difficulty and player engagement, operationalized as average pass and churn rates. We improve this approach by enhancing DRL with Monte Carlo Tree Search (MCTS). We also motivate an enhanced selection strategy for predictor features, based on the observation that an AI agent's best-case performance can yield stronger correlations with human data than the agent's average performance. Both additions consistently improve the prediction accuracy, and the DRL-enhanced MCTS outperforms both DRL and vanilla MCTS in the hardest levels. We conclude that player modelling via automated playtesting can benefit from combining DRL and MCTS. Moreover, it can be worthwhile to investigate a subset of repeated best AI agent runs, if AI gameplay does not yield good predictions on average.