Many real-world applications require artificial agents to compete and coordinate with other agents in complex environments. As a stepping stone to this goal, the domain of StarCraft has emerged by consensus as an important challenge for artificial intelligence research, owing to its iconic and enduring status among the most difficult professional esports and its relevance to the real world in terms of its raw complexity and multiagent challenges. Over the course of a decade and numerous competitions 1–3, the best results have been made possible by hand-crafting major elements of the system, simplifying important aspects of the game, or using superhuman capabilities 4. Even with these modifications, no previous system has come close to rivalling the skill of top players in the full game. We chose to address the challenge of StarCraft using general purpose learning methods that are in principle applicable to other complex domains: a multi-agent reinforcement learning algorithm that uses data from both human and agent games within a diverse league of continually adapting strategies and counterstrategies, each represented by deep neural networks5,6. We evaluated our agent, AlphaStar, in the full game of StarCraft II, through a series of online games against human players. AlphaStar was rated at Grandmaster level for all three StarCraft races and above 99.8% of officially ranked human players.

While deep reinforcement learning techniques have led to agents that are successfully able to learn to perform a number of tasks that had been previously unlearnable, these techniques are still susceptible to the longstanding problem of reward sparsity. This is especially true for tasks such as training an agent to play StarCraft II, a real-time strategy game where reward is only given at the end of a game which is usually very long. While this problem can be addressed through reward shaping, such approaches typically require a human expert with specialized knowledge. Inspired by the vision of enabling reward shaping through the more-accessible paradigm of natural-language narration, we develop a technique that can provide the benefits of reward shaping using natural language commands. Our narration-guided RL agent projects sequences of natural-language commands into the same high-dimensional representation space as corresponding goal states. We show that we can get improved performance with our method compared to traditional reward-shaping approaches. Additionally, we demonstrate the ability of our method to generalize to unseen natural-language commands.

Single-agent deep reinforcement learning has achieved impressive performance in many domains, including playing Go [1, 2] and Atari games [3, 4]. However, many real world problems, such as traffic congestion reduction [5, 6], antenna tilt control [7], and dynamic resource allocation [8] are more naturally modeled as multi-agent systems. Unfortunately, directly deploying single-agent reinforcement learning to each agent in a multi-agent system does not result in satisfying performance [9, 10]. Particularly, in multi-agent reinforcement learning [8, 10-19], estimating the value function is challenging, because the environment is non-stationary from the perspective of an individual agent [10, 11]. To alleviate the issue, recently, multi-agent deep deterministic policy gradient (MADDPG) [10] proposed a centralized critic whose input is the concatenation of all agents' observations and actions.

Exploration in environments with continuous control and sparse rewards remains a key challenge in reinforcement learning (RL). Recently, surprise has been used as an intrinsic reward that encourages systematic and efficient exploration. We introduce a new definition of surprise and its RL implementation named Variational Assorted Surprise Exploration (VASE). VASE uses a Bayesian neural network as a model of the environment dynamics and is trained using variational inference, alternately updating the accuracy of the agent's model and policy. Our experiments show that in continuous control sparse reward environments VASE outperforms other surprise-based exploration techniques.

User identity linkage is a task of recognizing the identities of the same user across different social networks (SN). Previous works tackle this problem via estimating the pairwise similarity between identities from different SN, predicting the label of identity pairs or selecting the most relevant identity pair based on the similarity scores. However, most of these methods ignore the results of previously matched identities, which could contribute to the linkage in following matching steps. To address this problem, we convert user identity linkage into a sequence decision problem and propose a reinforcement learning model to optimize the linkage strategy from the global perspective. Our method makes full use of both the social network structure and the history matched identities, and explores the long-term influence of current matching on subsequent decisions. We conduct experiments on different types of datasets, the results show that our method achieves better performance than other state-of-the-art methods.

Automatic machine learning (AutoML) is an area of research aimed at automating machine learning (ML) activities that currently require human experts. One of the most challenging tasks in this field is the automatic generation of end-to- end ML pipelines: combining multiple types of ML algorithms into a single architecture used for end-to-end analysis of previously-unseen data. This task has two challenging aspects: the first is the need to explore a large search space of algorithms and pipeline architectures. The second challenge is the computational cost of training and evaluating multiple pipelines. In this study we present DeepLine, a reinforcement learning based approach for automatic pipeline generation. Our proposed approach utilizes an efficient representation of the search space and leverages past knowledge gained from previously-analyzed datasets to make the problem more tractable. Additionally, we propose a novel hierarchical-actions algorithm that serves as a plugin, mediating the environment-agent interaction in deep reinforcement learning problems. The plugin significantly speeds up the training process of our model. Evaluation on 56 datasets shows that DeepLine outperforms state-of-the-art approaches both in accuracy and in computational cost.

Advancements in robotics and artificial intelligence (AI) are enabling the development of artificial agents designed to assist humans in a variety of everyday settings. One of the many possible uses for these systems could be to escort humans or valuable goods that are being transferred from one location to another, defending them from threats or attacks. Fascinated by this idea, a team of researchers at the University of New Mexico has recently introduced a new end-to-end solution for coordinating robotic escort teams that are protecting high-value payloads or goods. The technique they proposed, presented in a paper pre-published on arXiv, is based on deep reinforcement learning (RL), which entails training algorithms to make effective predictions by analyzing data. "I first came up with the idea behind this study when thinking about lugging my suitcase through a crowded airport," Lydia Tapia, the lead researcher on the study, told TechXplore.

$\varepsilon$-greedy is a policy used to balance exploration and exploitation in many reinforcement learning setting. In cases where the agent uses some on-policy algorithm to learn optimal behaviour, it makes sense for the agent to explore more initially and eventually exploit more as it approaches the target behaviour. This shift from heavy exploration to heavy exploitation can be represented as decay in the $\varepsilon$ value, where $\varepsilon$ depicts the how much an agent is allowed to explore. This paper proposes a new approach to this $\varepsilon$ decay where the decay is based on feedback from the environment. This paper also compares and contrasts one such approach based on rewards and compares it against standard exponential decay. The new approach, in the environments tested, produces more consistent results that on average perform better.

We introduce a framework for dynamic adversarial discovery of information (DADI), motivated by a scenario where information (a feature set) is used by third parties with unknown objectives. We train a reinforcement learning agent to sequentially acquire a subset of the information while balancing accuracy and fairness of predictors downstream. Based on the set of already acquired features, the agent decides dynamically to either collect more information from the set of available features or to stop and predict using the information that is currently available. Building on previous work exploring adversarial representation learning, we attain group fairness (demographic parity) by rewarding the agent with the adversary's loss, computed over the final feature set. Importantly, however, the framework provides a more general starting point for fair or private dynamic information discovery. Finally, we demonstrate empirically, using two real-world datasets, that we can trade-off fairness and predictive performance

Autonomous agents situated in real-world environments must be able to master large repertoires of skills. While a single short skill can be learned quickly, it would be impractical to learn every task independently. Instead, the agent should share knowledge across behaviors such that each task can be learned efficiently, and such that the resulting model can generalize to new tasks, especially ones that are compositions or subsets of tasks seen previously. A policy conditioned on a goal or demonstration has the potential to share knowledge between tasks if it sees enough diversity of inputs. However, these methods may not generalize to a more complex task at test time. We introduce compositional plan vectors (CPVs) to enable a policy to perform compositions of tasks without additional supervision. CPVs represent trajectories as the sum of the subtasks within them. We show that CPVs can be learned within a one-shot imitation learning framework without any additional supervision or information about task hierarchy, and enable a demonstration-conditioned policy to generalize to tasks that sequence twice as many skills as the tasks seen during training. Analogously to embeddings such as word2vec in NLP, CPVs can also support simple arithmetic operations -- for example, we can add the CPVs for two different tasks to command an agent to compose both tasks, without any additional training.