This paper studies a model for online job scheduling in green datacenters. In green datacenters, resource availability depends on the power supply from the renewables. Intermittent power supply from renewables leads to intermittent resource availability, inducing job delays (and associated costs). Green datacenter operators must intelligently manage their workloads and available power supply to extract maximum benefits. The scheduler's objective is to schedule jobs on a set of resources to maximize the total value (revenue) while minimizing the overall job delay. A trade-off exists between achieving high job value on the one hand and low expected delays on the other. Hence, the aims of achieving high rewards and low costs are in opposition. In addition, datacenter operators often prioritize multiple objectives, including high system utilization and job completion. To accomplish the opposing goals of maximizing total job value and minimizing job delays, we apply the Proportional-Integral-Derivative (PID) Lagrangian methods in Deep Reinforcement Learning to job scheduling problem in the green datacenter environment. Lagrangian methods are widely used algorithms for constrained optimization problems. We adopt a controls perspective to learn the Lagrange multiplier with proportional, integral, and derivative control, achieving favorable learning dynamics. Feedback control defines cost terms for the learning agent, monitors the cost limits during training, and continuously adjusts the learning parameters to achieve stable performance. Our experiments demonstrate improved performance compared to scheduling policies without the PID Lagrangian methods. Experimental results illustrate the effectiveness of the Constraint Controlled Reinforcement Learning (CoCoRL) scheduler that simultaneously satisfies multiple objectives.

In the future, powerful AI systems may be deployed in high-stakes settings, where a single failure could be catastrophic. One technique for improving AI safety in high-stakes settings is adversarial training, which uses an adversary to generate examples to train on in order to achieve better worst-case performance. In this work, we used a safe language generation task (``avoid injuries'') as a testbed for achieving high reliability through adversarial training. We created a series of adversarial training techniques -- including a tool that assists human adversaries -- to find and eliminate failures in a classifier that filters text completions suggested by a generator. In our task, we determined that we can set very conservative classifier thresholds without significantly impacting the quality of the filtered outputs. We found that adversarial training increased robustness to the adversarial attacks that we trained on -- doubling the time for our contractors to find adversarial examples both with our tool (from 13 to 26 minutes) and without (from 20 to 44 minutes) -- without affecting in-distribution performance. We hope to see further work in the high-stakes reliability setting, including more powerful tools for enhancing human adversaries and better ways to measure high levels of reliability, until we can confidently rule out the possibility of catastrophic deployment-time failures of powerful models.

Green Security Games with real-time information (GSG-I) add the real-time information about the agents' movement to the typical GSG formulation. Prior works on GSG-I have used deep reinforcement learning (DRL) to learn the best policy for the agent in such an environment without any need to store the huge number of state representations for GSG-I. However, the decision-making process of DRL methods is largely opaque, which results in a lack of trust in their predictions. To tackle this issue, we present an interpretable DRL method for GSG-I that generates visualization to explain the decisions taken by the DRL algorithm. We also show that this approach performs better and works well with a simpler training regimen compared to the existing method.

Development of navigation algorithms is essential for the successful deployment of robots in rapidly changing hazardous environments for which prior knowledge of configuration is often limited or unavailable. Use of traditional path-planning algorithms, which are based on localization and require detailed obstacle maps with goal locations, is not possible. In this regard, vision-based algorithms hold great promise, as visual information can be readily acquired by a robot's onboard sensors and provides a much richer source of information from which deep neural networks can extract complex patterns. Deep reinforcement learning has been used to achieve vision-based robot navigation. However, the efficacy of these algorithms in environments with dynamic obstacles and high variation in the configuration space has not been thoroughly investigated. In this paper, we employ a deep Dyna-Q learning algorithm for room evacuation and obstacle avoidance in partially observable environments based on low-resolution raw image data from an onboard camera. We explore the performance of a robotic agent in environments containing no obstacles, convex obstacles, and concave obstacles, both static and dynamic. Obstacles and the exit are initialized in random positions at the start of each episode of reinforcement learning. Overall, we show that our algorithm and training approach can generalize learning for collision-free evacuation of environments with complex obstacle configurations. It is evident that the agent can navigate to a goal location while avoiding multiple static and dynamic obstacles, and can escape from a concave obstacle while searching for and navigating to the exit.

Effectively exploring the environment is a key challenge in reinforcement learning (RL). We address this challenge by defining a novel intrinsic reward based on a foundation model, such as contrastive language image pretraining (CLIP), which can encode a wealth of domain-independent semantic visual-language knowledge about the world. Specifically, our intrinsic reward is defined based on pre-trained CLIP embeddings without any fine-tuning or learning on the target RL task. We demonstrate that CLIP-based intrinsic rewards can drive exploration towards semantically meaningful states and outperform state-of-the-art methods in challenging sparse-reward procedurally-generated environments.

Safe reinforcement learning (RL) studies problems where an intelligent agent has to not only maximize reward but also avoid exploring unsafe areas. In this study, we propose CUP, a novel policy optimization method based on Constrained Update Projection framework that enjoys rigorous safety guarantee. Central to our CUP development is the newly proposed surrogate functions along with the performance bound. Compared to previous safe RL methods, CUP enjoys the benefits of 1) CUP generalizes the surrogate functions to generalized advantage estimator (GAE), leading to strong empirical performance. 2) CUP unifies performance bounds, providing a better understanding and interpretability for some existing algorithms; 3) CUP provides a non-convex implementation via only first-order optimizers, which does not require any strong approximation on the convexity of the objectives. To validate our CUP method, we compared CUP against a comprehensive list of safe RL baselines on a wide range of tasks. Experiments show the effectiveness of CUP both in terms of reward and safety constraint satisfaction. We have opened the source code of CUP at this link https://github.com/zmsn-2077/ CUP-safe-rl.

New technologies and the availability of geospatial data have drawn attention to spatio-temporal biases present in society. For example: the COVID-19 pandemic highlighted disparities in the availability of broadband service and its role in the digital divide; the environmental justice movement in the United States has raised awareness to health implications for minority populations stemming from historical redlining practices; and studies have found varying quality and coverage in the collection and sharing of open-source geospatial data. Despite the extensive literature on machine learning (ML) fairness, few algorithmic strategies have been proposed to mitigate such biases. In this paper we highlight the unique challenges for quantifying and addressing spatio-temporal biases, through the lens of use cases presented in the scientific literature and media. We envision a roadmap of ML strategies that need to be developed or adapted to quantify and overcome these challenges -- including transfer learning, active learning, and reinforcement learning techniques. Further, we discuss the potential role of ML in providing guidance to policy makers on issues related to spatial fairness.

Text-based games present a unique class of sequential decision making problem in which agents interact with a partially observable, simulated environment via actions and observations conveyed through natural language. Such observations typically include instructions that, in a reinforcement learning (RL) setting, can directly or indirectly guide a player towards completing reward-worthy tasks. In this work, we study the ability of RL agents to follow such instructions. We conduct experiments that show that the performance of state-of-the-art text-based game agents is largely unaffected by the presence or absence of such instructions, and that these agents are typically unable to execute tasks to completion. To further study and address the task of instruction following, we equip RL agents with an internal structured representation of natural language instructions in the form of Linear Temporal Logic (LTL), a formal language that is increasingly used for temporally extended reward specification in RL. Our framework both supports and highlights the benefit of understanding the temporal semantics of instructions and in measuring progress towards achievement of such a temporally extended behaviour. Experiments with 500+ games in TextWorld demonstrate the superior performance of our approach.

We present state advantage weighting for offline reinforcement learning (RL). In contrast to action advantage $A(s,a)$ that we commonly adopt in QSA learning, we leverage state advantage $A(s,s^\prime)$ and QSS learning for offline RL, hence decoupling the action from values. We expect the agent can get to the high-reward state and the action is determined by how the agent can get to that corresponding state. Experiments on D4RL datasets show that our proposed method can achieve remarkable performance against the common baselines. Furthermore, our method shows good generalization capability when transferring from offline to online.

We study stochastic optimization algorithms for constrained nonconvex stochastic optimization problems with Markovian data. In particular, we focus on the case when the transition kernel of the Markov chain is state-dependent. Such stochastic optimization problems arise in various machine learning problems including strategic classification and reinforcement learning. For this problem, we study both projection-based and projection-free algorithms. In both cases, we establish that the number of calls to the stochastic first-order oracle to obtain an appropriately defined $\epsilon$-stationary point is of the order $\mathcal{O}(1/\epsilon^{2.5})$. In the projection-free setting we additionally establish that the number of calls to the linear minimization oracle is of order $\mathcal{O}(1/\epsilon^{5.5})$. We also empirically demonstrate the performance of our algorithm on the problem of strategic classification with neural networks.