Leave it to the folks at Google to devise AI capable of predicting which machine learning models will produce the best results. In a newly-published paper ("Off-Policy Evaluation via Off-Policy Classification") and blog post, a team of Google AI researchers propose what they call "off-policy classification," or OPC, which evaluates the performance of AI-driven agents by treating evaluation as a classification problem. The team notes that their approach -- a variant of reinforcement learning, which employs rewards to drive software policies toward goals -- works with image inputs and scales to tasks including vision-based robotic grasping. "Fully off-policy reinforcement learning is a variant in which an agent learns entirely from older data, which is appealing because it enables model iteration without requiring a physical robot," writes Robotics at Google software engineer Alexa Irpan. "With fully off-policy RL, one can train several models on the same fixed dataset collected by previous agents, then select the best one."

One of the primary goals of AI research is the development of artificial agents that can exceed human performance on a wide range of cognitive tasks, in other words, artificial general intelligence (AGI). Although the development of AGI has many potential benefits, there are also many safety concerns that have been raised in the literature [Bostrom, 2014; Everitt et al., 2018; Amodei et al., 2016]. Various approaches for addressing AGI safety have been proposed [Leike et al., 2018; Christiano et al., 2018; Irving et al., 2018; Hadfield-Menell et al., 2016; Everitt, 2018], often presented as a modification of the reinforcement learning (RL) framework, or a new framework altogether. Understanding and comparing different frameworks for AGI safety can be difficult because they build on differing concepts and assumptions. For example, both reward modeling [Leike et al., 2018] and cooperative inverse RL [Hadfield-Menell et al., 2016] are frameworks for making an agent learn the preferences of a human user, but what are the key differences between them?

In this paper, we study how to learn an appropriate lane changing strategy for autonomous vehicles by using deep reinforcement learning. We show that the reward of the system should consider the overall traffic efficiency instead of the travel efficiency of an individual vehicle. In summary, cooperation leads to a more harmonic and efficient traffic system rather than competition

Automatic music transcription is considered to be one of the hardest problems in music information retrieval, yet recent deep learning approaches have achieved substantial improvements on transcription performance. These approaches commonly employ supervised learning models that predict various time-frequency representations, by minimizing element-wise losses such as the cross entropy function. However, applying the loss in this manner assumes conditional independence of each label given the input, and thus cannot accurately express inter-label dependencies. To address this issue, we introduce an adversarial training scheme that operates directly on the time-frequency representations and makes the output distribution closer to the ground-truth. Through adversarial learning, we achieve a consistent improvement in both frame-level and note-level metrics over Onsets and Frames, a state-of-the-art music transcription model. Our results show that adversarial learning can significantly reduce the error rate while increasing the confidence of the model estimations. Our approach is generic and applicable to any transcription model based on multi-label predictions, which are very common in music signal analysis.

We present Placeto, a reinforcement learning (RL) approach to efficiently find device placements for distributed neural network training. Unlike prior approaches that only find a device placement for a specific computation graph, Placeto can learn generalizable device placement policies that can be applied to any graph. We propose two key ideas in our approach: (1) we represent the policy as performing iterative placement improvements, rather than outputting a placement in one shot; (2) we use graph embeddings to capture relevant information about the structure of the computation graph, without relying on node labels for indexing. These ideas allow Placeto to train efficiently and generalize to unseen graphs. Our experiments show that Placeto requires up to 6.1x fewer training steps to find placements that are on par with or better than the best placements found by prior approaches. Moreover, Placeto is able to learn a generalizable placement policy for any given family of graphs, which can then be used without any retraining to predict optimized placements for unseen graphs from the same family. This eliminates the large overhead incurred by prior RL approaches whose lack of generalizability necessitates re-training from scratch every time a new graph is to be placed.

Interference among concurrent transmissions in a wireless network is a key factor limiting the system performance. One way to alleviate this problem is to manage the radio resources in order to maximize either the average or the worst-case performance. However, joint consideration of both metrics is often neglected as they are competing in nature. In this article, a mechanism for radio resource management using multi-agent deep reinforcement learning (RL) is proposed, which strikes the right trade-off between maximizing the average and the $5^{th}$ percentile user throughput. Each transmitter in the network is equipped with a deep RL agent, receiving partial observations from the network (e.g., channel quality, interference level, etc.) and deciding whether to be active or inactive at each scheduling interval for given radio resources, a process referred to as link scheduling. Based on the actions of all agents, the network emits a reward to the agents, indicating how good their joint decisions were. The proposed framework enables the agents to make decisions in a distributed manner, and the reward is designed in such a way that the agents strive to guarantee a minimum performance, leading to a fair resource allocation among all users across the network. Simulation results demonstrate the superiority of our approach compared to decentralized baselines in terms of average and $5^{th}$ percentile user throughput, while achieving performance close to that of a centralized exhaustive search approach. Moreover, the proposed framework is robust to mismatches between training and testing scenarios. In particular, it is shown that an agent trained on a network with low transmitter density maintains its performance and outperforms the baselines when deployed in a network with a higher transmitter density.

Drawing an inspiration from behavioral studies of human decision making, we propose here a general parametric framework for a reinforcement learning problem, which extends the standard Q-learning approach to incorporate a two-stream framework of reward processing with biases biologically associated with several neurological and psychiatric conditions, including Parkinson's and Alzheimer's diseases, attention-deficit/hyperactivity disorder (ADHD), addiction, and chronic pain. For AI community, the development of agents that react differently to different types of rewards can enable us to understand a wide spectrum of multi-agent interactions in complex real-world socioeconomic systems. Moreover, from the behavioral modeling perspective, our parametric framework can be viewed as a first step towards a unifying computational model capturing reward processing abnormalities across multiple mental conditions and user preferences in long-term recommendation systems.

Our goal is to accurately and efficiently learn reward functions for autonomous robots. Current approaches to this problem include inverse reinforcement learning (IRL), which uses expert demonstrations, and preference-based learning, which iteratively queries the user for her preferences between trajectories. In robotics however, IRL often struggles because it is difficult to get high-quality demonstrations; conversely, preference-based learning is very inefficient since it attempts to learn a continuous, high-dimensional function from binary feedback. We propose a new framework for reward learning, DemPref, that uses both demonstrations and preference queries to learn a reward function. Specifically, we (1) use the demonstrations to learn a coarse prior over the space of reward functions, to reduce the effective size of the space from which queries are generated; and (2) use the demonstrations to ground the (active) query generation process, to improve the quality of the generated queries. Our method alleviates the efficiency issues faced by standard preference-based learning methods and does not exclusively depend on (possibly low-quality) demonstrations. In numerical experiments, we find that DemPref is significantly more efficient than a standard active preference-based learning method. In a user study, we compare our method to a standard IRL method; we find that users rated the robot trained with DemPref as being more successful at learning their desired behavior, and preferred to use the DemPref system (over IRL) to train the robot.

Reinforcement Learning (RL) of contact-rich manipulation tasks has yielded impressive results in recent years. While many studies in RL focus on varying the observation space or reward model, few efforts focused on the choice of action space (e.g. joint or end-effector space, position, velocity, etc.). However, studies in robot motion control indicate that choosing an action space that conforms to the characteristics of the task can simplify exploration and improve robustness to disturbances. This paper studies the effect of different action spaces in deep RL and advocates for Variable Impedance Control in End-effector Space (VICES) as an advantageous action space for constrained and contact-rich tasks. We evaluate multiple action spaces on three prototypical manipulation tasks: Path Following (task with no contact), Door Opening (task with kinematic constraints), and Surface Wiping (task with continuous contact). We show that VICES improves sample efficiency, maintains low energy consumption, and ensures safety across all three experimental setups. Further, RL policies learned with VICES can transfer across different robot models in simulation, and from simulation to real for the same robot. Further information is available at https://stanfordvl.github.io/vices.

Global routing has been a historically challenging problem in electronic circuit design, where the challenge is to connect a large and arbitrary number of circuit components with wires without violating the design rules for the printed circuit boards or integrated circuits. Similar routing problems also exist in the design of complex hydraulic systems, pipe systems and logistic networks. Existing solutions typically consist of greedy algorithms and hard-coded heuristics. As such, existing approaches suffer from a lack of model flexibility and non-optimum solutions. As an alternative approach, this work presents a deep reinforcement learning method for solving the global routing problem in a simulated environment. At the heart of the proposed method is deep reinforcement learning that enables an agent to produce an optimal policy for routing based on the variety of problems it is presented with leveraging the conjoint optimization mechanism of deep reinforcement learning. Conjoint optimization mechanism is explained and demonstrated in details; the best network structure and the parameters of the learned model are explored. Based on the fine-tuned model, routing solutions and rewards are presented and analyzed. The results indicate that the approach can outperform the benchmark method of a sequential A* method, suggesting a promising potential for deep reinforcement learning for global routing and other routing or path planning problems in general. Another major contribution of this work is the development of a global routing problem sets generator with the ability to generate parameterized global routing problem sets with different size and constraints, enabling evaluation of different routing algorithms and the generation of training datasets for future data-driven routing approaches.