Matthew Taylor

Taher Jafferjee, Juliusz Ziomek, Tianpei Yang et al. · oxford

Centralised training with decentralised execution (CT-DE) serves as the foundation of many leading multi-agent reinforcement learning (MARL) algorithms. Despite its popularity, it suffers from a critical drawback due to its reliance on learning from a single sample of the joint-action at a given state. As agents explore and update their policies during training, these single samples may poorly represent the actual joint-policy of the system of agents leading to high variance gradient estimates that hinder learning. To address this problem, we propose an enhancement tool that accommodates any actor-critic MARL method. Our framework, Performance Enhancing Reinforcement Learning Apparatus (PERLA), introduces a sampling technique of the agents' joint-policy into the critics while the agents train. This leads to TD updates that closely approximate the true expected value under the current joint-policy rather than estimates from a single sample of the joint-action at a given state. This produces low variance and precise estimates of expected returns, minimising the variance in the critic estimators which typically hinders learning. Moreover, as we demonstrate, by eliminating much of the critic variance from the single sampling of the joint policy, PERLA enables CT-DE methods to scale more efficiently with the number of agents. Theoretically, we prove that PERLA reduces variance in value estimates similar to that of decentralised training while maintaining the benefits of centralised training. Empirically, we demonstrate PERLA's superior performance and ability to reduce estimator variance in a range of benchmarks including Multi-agent Mujoco, and StarCraft II Multi-agent Challenge.

Daniel May, Matthew Taylor, Petr Musilek

As distributed energy resources (DERs) grow, the electricity grid faces increased net load variability at the grid edge, impacting operability and reliability. Transactive energy, facilitated through local energy markets, offers a decentralized, indirect demand response solution, with model-free control techniques, such as deep reinforcement learning (DRL), enabling automated, decentralized participation. However, existing studies largely overlook community-level net load variability, focusing instead on socioeconomic metrics. This study addresses this gap by using DRL agents to automate end-user participation in a local energy market (ALEX), where agents act independently to minimize individual energy bills. Results reveal a strong link between bill reduction and decreased net load variability, assessed across metrics such as ramping rate, load factor, and peak demand over various time horizons. Using a no-control baseline, DRL agents are benchmarked against a near-optimal dynamic programming approach. The dynamic programming benchmark achieves reductions of 22.05 percent, 83.92 percent, and 24.09 percent in daily import, export, and peak demand, respectively, while the DRL agents show comparable or superior results with reductions of 21.93 percent, 84.46 percent, and 27.02 percent. This study demonstrates the effectiveness of DRL in decentralized grid management, highlighting its scalability and near-optimal performance in reducing net load variability within community-driven energy markets.

Jerson Francia, Derek Hansen, Ben Schooley et al.

This paper explores the use of Large Language Models (LLMs) in spear phishing message generation and evaluates their performance compared to human-authored counterparts. Our pilot study examines the effectiveness of smishing (SMS phishing) messages created by GPT-4 and human authors, which have been personalized for willing targets. The targets assessed these messages in a modified ranked-order experiment using a novel methodology we call TRAPD (Threshold Ranking Approach for Personalized Deception). Experiments involved ranking each spear phishing message from most to least convincing, providing qualitative feedback, and guessing which messages were human- or AI-generated. Results show that LLM-generated messages are often perceived as more convincing than those authored by humans, particularly job-related messages. Targets also struggled to distinguish between human- and AI-generated messages. We analyze different criteria the targets used to assess the persuasiveness and source of messages. This study aims to highlight the urgent need for further research and improved countermeasures against personalized AI-enabled social engineering attacks.

David Mguni, Taher Jafferjee, Jianhong Wang et al.

Reward shaping (RS) is a powerful method in reinforcement learning (RL) for overcoming the problem of sparse or uninformative rewards. However, RS typically relies on manually engineered shaping-reward functions whose construction is time-consuming and error-prone. It also requires domain knowledge which runs contrary to the goal of autonomous learning. We introduce Reinforcement Learning Optimising Shaping Algorithm (ROSA), an automated reward shaping framework in which the shaping-reward function is constructed in a Markov game between two agents. A reward-shaping agent (Shaper) uses switching controls to determine which states to add shaping rewards for more efficient learning while the other agent (Controller) learns the optimal policy for the task using these shaped rewards. We prove that ROSA, which adopts existing RL algorithms, learns to construct a shaping-reward function that is beneficial to the task thus ensuring efficient convergence to high performance policies. We demonstrate ROSA's properties in three didactic experiments and show its superior performance against state-of-the-art RS algorithms in challenging sparse reward environments.

Matthew Taylor, Ruturaj K. Vaidya, Drew Davidson et al.

Package managers for software repositories based on a single programming language are very common. Examples include npm (JavaScript), and PyPI (Python). These tools encourage code reuse, making it trivial for developers to import external packages. Unfortunately, repositories' size and the ease with which packages can be published facilitates the practice of typosquatting: the uploading of a package with name similar to that of a highly popular package, typically with the aim of capturing some of the popular package's installs. Typosquatting has serious negative implications, resulting in developers importing malicious packages, or -- as we show -- code clones which do not incorporate recent security updates. In order to tackle this problem, we present SpellBound, a tool for identifying and reporting potentially erroneous imports to developers. SpellBound implements a novel typosquatting detection technique, based on an in-depth analysis of npm and PyPI. Our technique leverages a model of lexical similarity between names, and further incorporates the notion of package popularity. This approach flags cases where unknown/scarcely used packages would be installed in place of popular ones with similar names, before installation occurs. We evaluated SpellBound on both npm and PyPI, with encouraging results: SpellBound flags typosquatting cases while generating limited warnings (0.5% of total package installs), and low overhead (only 2.5% of package install time). Furthermore, SpellBound allowed us to confirm known cases of typosquatting and discover one high-profile, unknown case of typosquatting that resulted in a package takedown by the npm security team.

Weixun Wang, Jianye Hao, Yixi Wang et al.

The Iterated Prisoner's Dilemma has guided research on social dilemmas for decades. However, it distinguishes between only two atomic actions: cooperate and defect. In real-world prisoner's dilemmas, these choices are temporally extended and different strategies may correspond to sequences of actions, reflecting grades of cooperation. We introduce a Sequential Prisoner's Dilemma (SPD) game to better capture the aforementioned characteristics. In this work, we propose a deep multiagent reinforcement learning approach that investigates the evolution of mutual cooperation in SPD games. Our approach consists of two phases. The first phase is offline: it synthesizes policies with different cooperation degrees and then trains a cooperation degree detection network. The second phase is online: an agent adaptively selects its policy based on the detected degree of opponent cooperation. The effectiveness of our approach is demonstrated in two representative SPD 2D games: the Apple-Pear game and the Fruit Gathering game. Experimental results show that our strategy can avoid being exploited by exploitative opponents and achieve cooperation with cooperative opponents.

William Curran, Tim Brys, Matthew Taylor et al.

Reinforcement learning algorithms need to deal with the exponential growth of states and actions when exploring optimal control in high-dimensional spaces. This is known as the curse of dimensionality. By projecting the agent's state onto a low-dimensional manifold, we can represent the state space in a smaller and more efficient representation. By using this representation during learning, the agent can converge to a good policy much faster. We test this approach in the Mario Benchmarking Domain. When using dimensionality reduction in Mario, learning converges much faster to a good policy. But, there is a critical convergence-performance trade-off. By projecting onto a low-dimensional manifold, we are ignoring important data. In this paper, we explore this trade-off of convergence and performance. We find that learning in as few as 4 dimensions (instead of 9), we can improve performance past learning in the full dimensional space at a faster convergence rate.