We study query and computationally efficient planning algorithms with linear function approximation and a simulator. We assume that the agent only has local access to the simulator, meaning that the agent can only query the simulator at states that have been visited before. This setting is more practical than many prior works on reinforcement learning with a generative model. We propose an algorithm named confident Monte Carlo least square policy iteration (Confident MC-LSPI) for this setting. Under the assumption that the Q-functions of all deterministic policies are linear in known features of the state-action pairs, we show that our algorithm has polynomial query and computational complexities in the dimension of the features, the effective planning horizon and the targeted sub-optimality, while these complexities are independent of the size of the state space. One technical contribution of our work is the introduction of a novel proof technique that makes use of a virtual policy iteration algorithm. We use this method to leverage existing results on $\ell_\infty$-bounded approximate policy iteration to show that our algorithm can learn the optimal policy for the given initial state even only with local access to the simulator. We believe that this technique can be extended to broader settings beyond this work.

Sparse rewards and long time horizons remain challenging for reinforcement learning algorithms. Exploration bonuses can help in sparse reward settings by encouraging agents to explore the state space, while hierarchical approaches can assist with long-horizon tasks by decomposing lengthy tasks into shorter subtasks. We propose HAC Explore (HACx), a new method that combines these approaches by integrating the exploration bonus method Random Network Distillation (RND) into the hierarchical approach Hierarchical Actor-Critic (HAC). HACx outperforms either component method on its own, as well as an existing approach to combining hierarchy and exploration, in a set of difficult simulated robotics tasks. HACx is the first RL method to solve a sparse reward, continuous-control task that requires over 1,000 actions.

We use functional mirror ascent to propose a general framework (referred to as FMA-PG) for designing policy gradient methods. The functional perspective distinguishes between a policy's functional representation (what are its sufficient statistics) and its parameterization (how are these statistics represented) and naturally results in computationally efficient off-policy updates. For simple policy parameterizations, the FMA-PG framework ensures that the optimal policy is a fixed point of the updates. It also allows us to handle complex policy parameterizations (e.g., neural networks) while guaranteeing policy improvement. Our framework unifies several PG methods and opens the way for designing sample-efficient variants of existing methods. Moreover, it recovers important implementation heuristics (e.g., using forward vs reverse KL divergence) in a principled way. With a softmax functional representation, FMA-PG results in a variant of TRPO with additional desirable properties. It also suggests an improved variant of PPO, whose robustness and efficiency we empirically demonstrate on MuJoCo. Via experiments on simple reinforcement learning problems, we evaluate algorithms instantiated by FMA-PG.

Machine learning (ML) has been strategic to Amazon from the early years. We are pioneers in areas such as recommendation engines, product search, eCommerce fraud detection, and large-scale optimization of fulfillment center operations. The Amazon ML Solutions Lab team helps AWS customers accelerate the use of machine learning to solve business and operational challenges and promote innovation in their organization. As a science manager for the ML Solutions Lab team, you will lead a team of customer-facing scientists and architects to design and deliver advanced ML solutions to solve diverse real-world problems for customers across all industries. You will interact with customers, translate their business problems into ML problems, and lead your team in applying classical ML algorithms and cutting-edge deep learning (DL) and reinforcement learning approaches to areas such as drug discovery, customer segmentation, fraud prevention, capacity planning, predictive maintenance, pricing optimization, call center analytics, player pose estimation, event detection, and virtual assistant among others.

In the age of big data, data processing and analytics are fundamental, ubiquitous, and crucial to many organizations which undertake a digitalization journey to improve and transform their businesses and operations. Data analytics typically entails other key operations such as data acquisition, data cleansing, data integration, modeling, etc., before insights could be extracted. Big data can unleash significant value creation across many sectors such as health care and retail[56]. However, the complexity of data (e.g., high volume, high velocity, and high variety) presents many challenges in data analytics and hence renders the difficulty in drawing meaningful insights. To tackle the challenge and facilitate the data processing and analytics efficiently and effectively, a lot of algorithms and techniques have been designed and numerous learning systems have also been developed by researchers and practitioners such as Spark MLlib[63], and Rafiki[104]. To support fast data processing and accurate data analytics, a huge number of algorithms rely on rules that are developed based on human knowledge and experience. For example, Shortest-job-first is a scheduling algorithm that chooses the job with the smallest execution time for the next execution. However, without fully exploiting characteristics of the workload, it can achieve inferior performance compared to DRL-based scheduling algorithm [58].

Autonomous driving in multi-agent and dynamic traffic scenarios is challenging, where the behaviors of other road agents are uncertain and hard to model explicitly, and the ego-vehicle should apply complicated negotiation skills with them to achieve both safe and efficient driving in various settings, such as giving way, merging and taking turns. Traditional planning methods are largely rule-based and scale poorly in these complex dynamic scenarios, often leading to reactive or even overly conservative behaviors. Therefore, they require tedious human efforts to maintain workability. Recently, deep learning-based methods have shown promising results with better generalization capability but less hand engineering effort. However, they are either implemented with supervised imitation learning (IL) that suffers from the dataset bias and distribution mismatch problems, or trained with deep reinforcement learning (DRL) but focus on one specific traffic scenario. In this work, we propose DQ-GAT to achieve scalable and proactive autonomous driving, where graph attention-based networks are used to implicitly model interactions, and asynchronous deep Q-learning is employed to train the network end-to-end in an unsupervised manner. Extensive experiments through a high-fidelity driving simulation show that our method can better trade-off safety and efficiency in both seen and unseen scenarios, achieving higher goal success rates than the baselines (at most 4.7$\times$) with comparable task completion time. Demonstration videos are available at https://caipeide.github.io/dq-gat/.

Reinforcement learning (RL) is successful at learning to play games where the entire environment is visible. However, RL approaches are challenged in complex games like Starcraft II and in real-world environments where the entire environment is not visible. In these more complex games with more limited visual information, agents must choose where to look and how to optimally use their limited visual information in order to succeed at the game. We verify that with a relatively simple model the agent can learn where to look in scenarios with a limited visual bandwidth. We develop a method for masking part of the environment in Atari games to force the RL agent to learn both where to look and how to play the game in order to study where the RL agent learns to look. In addition, we develop a neural network architecture and method for allowing the agent to choose where to look and what action to take in the Pong game. Further, we analyze the strategies the agent learns to better understand how the RL agent learns to play the game.

Recently, needs for unmanned aerial vehicles (UAVs) that are attachable to the wall have been highlighted. As one of the ways to address the need, researches on various tilting multirotors that can increase maneuverability has been employed. Unfortunately, existing studies on the tilting multirotors require considerable amounts of prior information on the complex dynamic model. Meanwhile, reinforcement learning on quadrotors has been studied to mitigate this issue. Yet, these are only been applied to standard quadrotors, whose systems are less complex than those of tilting multirotors. In this paper, a novel reinforcement learning-based method is proposed to control a tilting multirotor on real-world applications, which is the first attempt to apply reinforcement learning to a tilting multirotor. To do so, we propose a novel reward function for a neural network model that takes power efficiency into account. The model is initially trained over a simulated environment and then fine-tuned using real-world data in order to overcome the sim-to-real gap issue. Furthermore, a novel, efficient state representation with respect to the goal frame that helps the network learn optimal policy better is proposed. As verified on real-world experiments, our proposed method shows robust controllability by overcoming the complex dynamics of tilting multirotors.

Typically, Reinforcement Learning (RL) assumes access to a pre-specified reward and then learns a policy maximizing the expected average of this reward along a trajectory. However, specifying rewards is difficult for many practical tasks (Atkeson & Schaal, 1997; Zhang et al., 2018; Ibarz et al., 2018). In such cases, it is convenient to instead perform Imitation Learning (IL), learning a policy from expert demonstrations. There are two major categories of Imitation Learning algorithms: Behavioral Cloning and Inverse Reinforcement Learning. Behavioral Cloning learns the policy by supervised learning on expert data, but is not robust to training errors, failing in settings where expert data is limited (Ross & Bagnell, 2010). Inverse Reinforcement Learning (IRL) achieves improved performance on limited data by constructing reward signals and calling an RL oracle to maximize these rewards (Ng et al., 2000).

As the field of machine learning for combinatorial optimization advances, traditional problems are resurfaced and readdressed through this new perspective. The overwhelming majority of the literature focuses on small graph problems, while several real-world problems are devoted to large graphs. Here, we focus on two such problems that are related: influence estimation, a \#P-hard counting problem, and influence maximization, an NP-hard problem. We develop GLIE, a Graph Neural Network (GNN) that inherently parameterizes an upper bound of influence estimation and train it on small simulated graphs. Experiments show that GLIE can provide accurate predictions faster than the alternatives for graphs 10 times larger than the train set. More importantly, it can be used on arbitrary large graphs for influence maximization, as the predictions can rank effectively seed sets even when the accuracy deteriorates. To showcase this, we propose a version of a standard Influence Maximization (IM) algorithm where we substitute traditional influence estimation with the predictions of GLIE.We also transfer GLIE into a reinforcement learning model that learns how to choose seeds to maximize influence sequentially using GLIE's hidden representations and predictions. The final results show that the proposed methods surpasses a previous GNN-RL approach and perform on par with a state-of-the-art IM algorithm.