Simon Khan

Alexander Grushin, Walt Woods, Alvaro Velasquez et al.

Any autonomous controller will be unsafe in some situations. The ability to quantitatively identify when these unsafe situations are about to occur is crucial for drawing timely human oversight in, e.g., freight transportation applications. In this work, we demonstrate that the true criticality of an agent's situation can be robustly defined as the mean reduction in reward given some number of random actions. Proxy criticality metrics that are computable in real-time (i.e., without actually simulating the effects of random actions) can be compared to the true criticality, and we show how to leverage these proxy metrics to generate safety margins, which directly tie the consequences of potentially incorrect actions to an anticipated loss in overall performance. We evaluate our approach on learned policies from APE-X and A3C within an Atari environment, and demonstrate how safety margins decrease as agents approach failure states. The integration of safety margins into programs for monitoring deployed agents allows for the real-time identification of potentially catastrophic situations.

Walt Woods, Alexander Grushin, Simon Khan et al.

AI-enabled capabilities are reaching the requisite level of maturity to be deployed in the real world, yet do not always make correct or safe decisions. One way of addressing these concerns is to leverage AI control systems alongside and in support of human decisions, relying on the AI control system in safe situations while calling on a human co-decider for critical situations. We extend a methodology for adversarial explanations (AE) to state-of-the-art reinforcement learning frameworks, including MuZero. Multiple improvements to the base agent architecture are proposed. We demonstrate how this technology has two applications: for intelligent decision tools and to enhance training / learning frameworks. In a decision support context, adversarial explanations help a user make the correct decision by highlighting those contextual factors that would need to change for a different AI-recommended decision. As another benefit of adversarial explanations, we show that the learned AI control system demonstrates robustness against adversarial tampering. Additionally, we supplement AE by introducing strategically similar autoencoders (SSAs) to help users identify and understand all salient factors being considered by the AI system. In a training / learning framework, this technology can improve both the AI's decisions and explanations through human interaction. Finally, to identify when AI decisions would most benefit from human oversight, we tie this combined system to our prior art on statistically verified analyses of the criticality of decisions at any point in time.

Alexander Grushin, Walt Woods, Alvaro Velasquez et al.

State of the art reinforcement learning methods sometimes encounter unsafe situations. Identifying when these situations occur is of interest both for post-hoc analysis and during deployment, where it might be advantageous to call out to a human overseer for help. Efforts to gauge the criticality of different points in time have been developed, but their accuracy is not well established due to a lack of ground truth, and they are not designed to be easily interpretable by end users. Therefore, we seek to define a criticality framework with both a quantifiable ground truth and a clear significance to users. We introduce true criticality as the expected drop in reward when an agent deviates from its policy for n consecutive random actions. We also introduce the concept of proxy criticality, a low-overhead metric that has a statistically monotonic relationship to true criticality. Safety margins make these interpretable, when defined as the number of random actions for which performance loss will not exceed some tolerance with high confidence. We demonstrate this approach in several environment-agent combinations; for an A3C agent in an Atari Beamrider environment, the lowest 5% of safety margins contain 47% of agent losses; i.e., supervising only 5% of decisions could potentially prevent roughly half of an agent's errors. This criticality framework measures the potential impacts of bad decisions, even before those decisions are made, allowing for more effective debugging and oversight of autonomous agents.

Qinwei Huang, Rui Zuo, Simon Khan et al.

Conventional federated learning assumes that greater learner participation improves training performance, by leveraging abundant, independently generated local data. However, in federated reinforcement learning (FRL) for unmanned aerial vehicle (UAV) teams in hazardous environments where experience generation is severely constrained by safety considerations, energy limitations, and mission duration, this assumption may break. This work introduces Experience-Constrained Hierarchical Federated Reinforcement Learning (EC-HFRL), a framework in which clusters act as federated learning agents, while multiple intra-cluster learners represent parallel learning resources that reuse a shared experience pool. We show that increasing participation does not necessarily improve learning performance. Instead, learning performance is strongly associated with experience reuse strategy and the dominance of key analytically identified gradient transition experiences within a cluster. In particular, minibatch size primarily determines effective replay exposure, while higher intra-cluster participation increases reuse level. Empirical results demonstrate that the performance regimes are strongly associated with the structure of the learning signal, rather than federated aggregation effects, clarifying the limited and secondary role of learner participation in experience-constrained FRL.

Sanup S. Araballi, Simon Khan, Chilukuri K. Mohan

Deep Reinforcement Learning (DRL) agents achieve remarkable performance in continuous control but remain opaque, hindering deployment in safety-critical domains. Existing explainability methods either provide only local insights (SHAP, LIME) or employ over-simplified surrogates failing to capture continuous dynamics (decision trees). This work proposes a Hierarchical Takagi-Sugeno-Kang (TSK) Fuzzy Classifier System (FCS) distilling neural policies into human-readable IF-THEN rules through K-Means clustering for state partitioning and Ridge Regression for local action inference. Three quantifiable metrics are introduced: Fuzzy Rule Activation Density (FRAD) measuring explanation focus, Fuzzy Set Coverage (FSC) validating vocabulary completeness, and Action Space Granularity (ASG) assessing control mode diversity. Dynamic Time Warping (DTW) validates temporal behavioral fidelity. Empirical evaluation on \textit{Lunar Lander(Continuous)} shows the Triangular membership function variant achieves 81.48\% $\pm$ 0.43\% fidelity, outperforming Decision Trees by 21 percentage points. The framework exhibits statistically superior interpretability (FRAD = 0.814 vs. 0.723 for Gaussian, $p < 0.001$) with low MSE (0.0053) and DTW distance (1.05). Extracted rules such as ``IF lander drifting left at high altitude THEN apply upward thrust with rightward correction'' enable human verification, establishing a pathway toward trustworthy autonomous systems.

Zilong Deng, Simon Khan, Shaofeng Zou

In this work, we study the sample complexity problem of risk-sensitive Reinforcement Learning (RL) with a generative model, where we aim to maximize the Conditional Value at Risk (CVaR) with risk tolerance level $τ$ at each step, a criterion we refer to as Iterated CVaR. We first build a connection between Iterated CVaR RL and $(s, a)$-rectangular distributional robust RL with a specific uncertainty set for CVaR. We establish nearly matching upper and lower bounds on the sample complexity of this problem. Specifically, we first prove that a value iteration-based algorithm, ICVaR-VI, achieves an $ε$-optimal policy with at most $\tilde{O} \left(\frac{SA}{(1-γ)^4τ^2ε^2} \right)$ samples, where $γ$ is the discount factor, and $S, A$ are the sizes of the state and action spaces. Furthermore, when $τ\geq γ$, the sample complexity improves to $\tilde{O} \left( \frac{SA}{(1-γ)^3ε^2} \right)$. We further show a minimax lower bound of $\tilde{O} \left(\frac{(1-γτ)SA}{(1-γ)^4τε^2} \right)$. For a fixed risk level $τ\in (0,1]$, our upper and lower bounds match, demonstrating the tightness and optimality of our analysis. We also investigate a limiting case with a small risk level $τ$, called Worst-Path RL, where the objective is to maximize the minimum possible cumulative reward. We develop matching upper and lower bounds of $\tilde{O} \left(\frac{SA}{p_{\min}} \right)$, where $p_{\min}$ denotes the minimum non-zero reaching probability of the transition kernel.

Ruipeng Liu, Qinru Qiu, Simon Khan et al.

A computational bottleneck in current Vector-Symbolic Architectures (VSAs) is the ``clean-up'' step, which decodes the noisy vectors retrieved from the architecture. Clean-up typically compares noisy vectors against a ``codebook'' of prototype vectors, incurring computational complexity that is quadratic or similar. We present a new codebook representation that supports efficient clean-up, based on Kroneker products of rotation-like matrices. The resulting clean-up time complexity is linearithmic, i.e. $\mathcal{O}(N\,\text{log}\,N)$, where $N$ is the vector dimension and also the number of vectors in the codebook. Clean-up space complexity is $\mathcal{O}(N)$. Furthermore, the codebook is not stored explicitly in computer memory: It can be represented in $\mathcal{O}(\text{log}\,N)$ space, and individual vectors in the codebook can be materialized in $\mathcal{O}(N)$ time and space. At the same time, asymptotic memory capacity remains comparable to standard approaches. Computer experiments confirm these results, demonstrating several orders of magnitude more scalability than baseline VSA techniques.

Timothy K. Mathes, Jessica Inman, Andrés Colón et al.

Despite the impressive feats demonstrated by Reinforcement Learning (RL), these algorithms have seen little adoption in high-risk, real-world applications due to current difficulties in explaining RL agent actions and building user trust. We present Counterfactual Demonstrations for Explanation (CODEX), a method that incorporates semantic clustering, which can effectively summarize RL agent behavior in the state-action space. Experimentation on the MiniGrid and StarCraft II gaming environments reveals the semantic clusters retain temporal as well as entity information, which is reflected in the constructed summary of agent behavior. Furthermore, clustering the discrete+continuous game-state latent representations identifies the most crucial episodic events, demonstrating a relationship between the latent and semantic spaces. This work contributes to the growing body of work that strives to unlock the power of RL for widespread use by leveraging and extending techniques from Natural Language Processing.

Rui Zuo, Simon Khan, Zifan Wang et al.

Reinforcement learning (RL) has demonstrated remarkable success in solving complex decision-making problems, yet its adoption in critical domains is hindered by the lack of interpretability in its decision-making processes. Existing explainable AI (xAI) approaches often fail to provide meaningful explanations for RL agents, particularly because they overlook the contrastive nature of human reasoning--answering "why this action instead of that one?". To address this gap, we propose a novel framework of contrastive learning to explain RL selected actions, named $\textbf{VisionMask}$. VisionMask is trained to generate explanations by explicitly contrasting the agent's chosen action with alternative actions in a given state using a self-supervised manner. We demonstrate the efficacy of our method through experiments across diverse RL environments, evaluating it in terms of faithfulness, robustness, and complexity. Our results show that VisionMask significantly improves human understanding of agent behavior while maintaining accuracy and fidelity. Furthermore, we present examples illustrating how VisionMask can be used for counterfactual analysis. This work bridges the gap between RL and xAI, paving the way for safer and more interpretable RL systems.

Qinwei Huang, Chen Luo, Alex B. Wu et al.

In a multi-agent system, agents share their local observations to gain global situational awareness for decision making and collaboration using a message passing system. When to send a message, how to encode a message, and how to leverage the received messages directly affect the effectiveness of the collaboration among agents. When training a multi-agent cooperative game using reinforcement learning (RL), the message passing system needs to be optimized together with the agent policies. This consequently increases the model's complexity and poses significant challenges to the convergence and performance of learning. To address this issue, we propose the Belief-map Assisted Multi-agent System (BAMS), which leverages a neuro-symbolic belief map to enhance training. The belief map decodes the agent's hidden state to provide a symbolic representation of the agent's understanding of the environment and other agent's status. The simplicity of symbolic representation allows the gathering and comparison of the ground truth information with the belief, which provides an additional channel of feedback for the learning. Compared to the sporadic and delayed feedback coming from the reward in RL, the feedback from the belief map is more consistent and reliable. Agents using BAMS can learn a more effective message passing network to better understand each other, resulting in better performance in a cooperative predator and prey game with varying levels of map complexity and compare it to previous multi-agent message passing models. The simulation results showed that BAMS reduced training epochs by 66\%, and agents who apply the BAMS model completed the game with 34.62\% fewer steps on average.

Shuang Ao, Simon Khan, Haris Aziz et al.

Understanding the agent's learning process, particularly the factors that contribute to its success or failure post-training, is crucial for comprehending the rationale behind the agent's decision-making process. Prior methods clarify the learning process by creating a structural causal model (SCM) or visually representing the distribution of value functions. Nevertheless, these approaches have constraints as they exclusively function in 2D-environments or with uncomplicated transition dynamics. Understanding the agent's learning process in complicated environments or tasks is more challenging. In this paper, we propose REVEAL-IT, a novel framework for explaining the learning process of an agent in complex environments. Initially, we visualize the policy structure and the agent's learning process for various training tasks. By visualizing these findings, we can understand how much a particular training task or stage affects the agent's performance in test. Then, a GNN-based explainer learns to highlight the most important section of the policy, providing a more clear and robust explanation of the agent's learning process. The experiments demonstrate that explanations derived from this framework can effectively help in the optimization of the training tasks, resulting in improved learning efficiency and final performance.