Author name cluster

Usman Anwar

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

10 papers

2 author rows

JBHI Journal 2026 Journal Article

Multimodal Cognitive Load Estimation With Radio Frequency Sensing and Pupillometry in Complex Auditory Environments

Usman Anwar
Adeel Hussain
Mandar Gogate
Kia Dashtipour
Tughrul Arslan
Amir Hussain
Peter Lomax

The detection of listening effort or cognitive load (CL) has been a major research challenge in recent years. Most conventional techniques utilise physiological or audio-visual sensors and are privacy-invasive and computationally complex. The challenges of synchronization, data alignment and accessibility limitations potentially increase the noise and error probability, compromising the accuracy of CL estimates. This innovative work presents a multi-modal, non-invasive and privacy-preserving approach that combines Radio Frequency (RF) and pupillometry sensing to address these challenges. Custom RF sensors are first designed and developed to capture blood flow changes in specific brain regions with high spatial resolution. Next, multi-modal fusion with pupillometry sensing is proposed and shown to offer a robust assessment of cognitive and listening effort through pupil size and pupil dilation. Our novel approach evaluates RF sensing to estimate CL from cerebral blood flow variations utilizing pupillometry as a baseline. A first-of-its-kind, multi-modal dataset is collected as a new benchmark resource in a controlled environment with participants to comprehend target speech with varying background noise levels. The framework is statistically evaluated using intraclass correlation for pupillometry data (average ICC> 0. 95). The correlation between pupillometry and RF data is established through Pearson's correlation (average PCC> 0. 79). Further, CL is classified into high and low categories based on RF data using K-means clustering. Future work involves integrating RF sensors with glasses to estimate listening effort for hearing-aid users and utilising RF measurements to optimize speech enhancement based on individual's listening effort and complexity of acoustic environment.

Details DOI

ICLR Conference 2025 Conference Paper

Interpreting Emergent Planning in Model-Free Reinforcement Learning

Thomas Bush
Stephen Chung
Usman Anwar
Adrià Garriga-Alonso
David Krueger 0001

We present the first mechanistic evidence that model-free reinforcement learning agents can learn to plan. This is achieved by applying a methodology based on concept-based interpretability to a model-free agent in Sokoban -- a commonly used benchmark for studying planning. Specifically, we demonstrate that DRC, a generic model-free agent introduced by [Guez et al. (2019)](https://arxiv.org/abs/1901.03559), uses learned concept representations to internally formulate plans that both predict the long-term effects of actions on the environment and influence action selection. Our methodology involves: (1) probing for planning-relevant concepts, (2) investigating plan formation within the agent's representations, and (3) verifying that discovered plans (in the agent's representations) have a causal effect on the agent's behavior through interventions. We also show that the emergence of these plans coincides with the emergence of a planning-like property: the ability to benefit from additional test-time compute. Finally, we perform a qualitative analysis of the planning algorithm learned by the agent and discover a strong resemblance to parallelized bidirectional search. Our findings advance understanding of the internal mechanisms underlying planning behavior in agents, which is important given the recent trend of emergent planning and reasoning capabilities in LLMs through RL.

Details

RLJ Journal 2025 Journal Article

Mitigating Goal Misgeneralization via Minimax Regret

Karim Abdel Sadek
Matthew Farrugia-Roberts
Usman Anwar
Hannah Erlebach
Christian Schroeder de Witt
David Krueger
Michael D Dennis

Safe generalization in reinforcement learning requires not only that a learned policy acts capably in new situations, but also that it uses its capabilities towards the pursuit of the designer’s intended goal. The latter requirement may fail when a proxy goal incentivizes similar behavior to the intended goal within the training environment, but not in novel deployment environments. This creates the risk that policies will behave as if in pursuit of the proxy goal, rather than the intended goal, in deployment—a phenomenon known as goal misgeneralization. In this paper, we formalize this problem setting in order to theoretically study the possibility of goal misgeneralization under different training objectives. We show that goal misgeneralization is possible under approximate optimization of the maximum expected value (MEV) objective, but not the minimax expected regret (MMER) objective. We then empirically show that the standard MEV-based training method of domain randomization exhibits goal misgeneralization in procedurally-generated grid-world environments, whereas current regret-based unsupervised environment design (UED) methods are more robust to goal misgeneralization (though they don’t find MMER policies in all cases). Our findings suggest that minimax expected regret is a promising approach to mitigating goal misgeneralization.

PDF Details

RLC Conference 2025 Conference Paper

Mitigating Goal Misgeneralization via Minimax Regret

Karim Abdel Sadek
Matthew Farrugia-Roberts
Usman Anwar
Hannah Erlebach
Christian Schroeder de Witt
David Krueger
Michael D Dennis

PDF Details

TMLR Journal 2025 Journal Article

Understanding In-Context Learning of Linear Models in Transformers Through an Adversarial Lens

Usman Anwar
Johannes von Oswald
Louis Kirsch
David Krueger
Spencer Frei

In this work, we make two contributions towards understanding of in-context learning of linear models by transformers. First, we investigate the adversarial robustness of in-context learning in transformers to hijacking attacks — a type of adversarial attacks in which the adversary’s goal is to manipulate the prompt to force the transformer to generate a specific output. We show that both linear transformers and transformers with GPT-2 architectures are vulnerable to such hijacking attacks. However, adversarial robustness to such attacks can be significantly improved through adversarial training --- done either at the pretraining or finetuning stage --- and can generalize to stronger attack models. Our second main contribution is a comparative analysis of adversarial vulnerabilities across transformer models and other algorithms for learning linear models. This reveals two novel findings. First, adversarial attacks transfer poorly between larger transformer models trained from different seeds despite achieving similar in-distribution performance. This suggests that transformers of the same architecture trained according to the same recipe may implement different in-context learning algorithms for the same task. Second, we observe that attacks do not transfer well between classical learning algorithms for linear models (single-step gradient descent and ordinary least squares) and transformers. This suggests that there could be qualitative differences between the in-context learning algorithms that transformers implement and these traditional algorithms.

PDF Details

TMLR Journal 2024 Journal Article

Foundational Challenges in Assuring Alignment and Safety of Large Language Models

Usman Anwar
Abulhair Saparov
Javier Rando
Daniel Paleka
Miles Turpin
Peter Hase
Ekdeep Singh Lubana
Erik Jenner

This work identifies 18 foundational challenges in assuring the alignment and safety of large language models (LLMs). These challenges are organized into three different categories: scientific understanding of LLMs, development and deployment methods, and sociotechnical challenges. Based on the identified challenges, we pose 200+, concrete research questions.

PDF Details

ICLR Conference 2024 Conference Paper

Reward Model Ensembles Help Mitigate Overoptimization

Thomas Coste
Usman Anwar
Robert Kirk
David Krueger 0001

Reinforcement learning from human feedback (RLHF) is a standard approach for fine-tuning large language models to follow instructions. As part of this process, learned reward models are used to approximately model human preferences. However, as imperfect representations of the “true” reward, these learned reward models are susceptible to overoptimization. Gao et al. (2023) studied this phenomenon in a synthetic human feedback setup with a significantly larger “gold” reward model acting as the true reward (instead of humans) and showed that overoptimization remains a persistent problem regardless of the size of the proxy reward model and training data used. Using a similar setup, we conduct a systematic study to evaluate the efficacy of using ensemble-based conservative optimization objectives, specifically worst-case optimization (WCO) and uncertainty-weighted optimization (UWO), for mitigating reward model overoptimization when using two optimization methods: (a) best-of-n sampling (BoN) (b) proximal policy optimization (PPO). We additionally extend the setup of Gao et al. (2023) to include 25% label noise to better mirror real-world conditions. Both with and without label noise we find that conservative optimization practically eliminates overoptimization and improves performance by up to 70% for BoN sampling. For PPO, ensemble-based conservative optimization always reduces overoptimization and outperforms single reward model optimization. Moreover, combining it with a small KL penalty successfully prevents overoptimization at no performance cost. Overall, our results demonstrate that ensemble-based conservative optimization can effectively counter overoptimization.

Details

TMLR Journal 2023 Journal Article

Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback

Stephen Casper
Xander Davies
Claudia Shi
Thomas Krendl Gilbert
Jérémy Scheurer
Javier Rando
Rachel Freedman
Tomek Korbak

Reinforcement learning from human feedback (RLHF) is a technique for training AI systems to align with human goals. RLHF has emerged as the central method used to finetune state-of-the-art large language models (LLMs). Despite this popularity, there has been relatively little public work systematizing its flaws. In this paper, we (1) survey open problems and fundamental limitations of RLHF and related methods; (2) overview techniques to understand, improve, and complement RLHF in practice; and (3) propose auditing and disclosure standards to improve societal oversight of RLHF systems. Our work emphasizes the limitations of RLHF and highlights the importance of a multi-layered approach to the development of safer AI systems.

PDF Details

TMLR Journal 2022 Journal Article

Bayesian Methods for Constraint Inference in Reinforcement Learning

Dimitris Papadimitriou
Usman Anwar
Daniel S. Brown

Learning constraints from demonstrations provides a natural and efficient way to improve the safety of AI systems; however, prior work only considers learning a single, point-estimate of the constraints. By contrast, we consider the problem of inferring constraints from demonstrations using a Bayesian perspective. We propose Bayesian Inverse Constraint Reinforcement Learning (BICRL), a novel approach that infers a posterior probability distribution over constraints from demonstrated trajectories. The main advantages of BICRL, compared to prior constraint inference algorithms, are (1) the freedom to infer constraints from partial trajectories and even from disjoint state-action pairs, (2) the ability to infer constraints from suboptimal demonstrations and in stochastic environments, and (3) the opportunity to use the posterior distribution over constraints in order to implement active learning and robust policy optimization techniques. We show that BICRL outperforms pre-existing constraint learning approaches, leading to more accurate constraint inference and consequently safer policies. We further propose Hierarchical BICRL that infers constraints locally in sub-spaces of the entire domain and then composes global constraint estimates leading to accurate and computationally efficient constraint estimation.

PDF Details

ICML Conference 2021 Conference Paper

Inverse Constrained Reinforcement Learning

Shehryar Malik
Usman Anwar
Alireza Aghasi
Ali Ahmed 0004

In real world settings, numerous constraints are present which are hard to specify mathematically. However, for the real world deployment of reinforcement learning (RL), it is critical that RL agents are aware of these constraints, so that they can act safely. In this work, we consider the problem of learning constraints from demonstrations of a constraint-abiding agent’s behavior. We experimentally validate our approach and show that our framework can successfully learn the most likely constraints that the agent respects. We further show that these learned constraints are \textit{transferable} to new agents that may have different morphologies and/or reward functions. Previous works in this regard have either mainly been restricted to tabular (discrete) settings, specific types of constraints or assume the environment’s transition dynamics. In contrast, our framework is able to learn arbitrary \textit{Markovian} constraints in high-dimensions in a completely model-free setting. The code is available at: \url{https: //github. com/shehryar-malik/icrl}.

Details