Search

Reinforcement learning is a powerful model of animal learning in brief, controlled experimental conditions, but does not readily explain the development of behavior over an animal's whole lifetime. In this paper, we describe a framework to address this shortcoming by introducing the single-life reinforcement learning setting to cognitive science. We construct an agent with two learning systems: an extrinsic learner that learns within a single lifetime, and an intrinsic learner that learns across lifetimes, equipping the agent with intrinsic motivation. We show that this model outperforms heuristic benchmarks and recapitulates a transition from exploratory to habit-driven behavior, while allowing the agent to learn an interpretable value function. We formulate a precise definition of intrinsic motivation and discuss the philosophical implications of using reinforcement learning as a model of behavior in the real world.

Recent capability increases in large language models (LLMs) open up applications in which teams of communicating generative AI agents solve joint tasks. This poses privacy and security challenges concerning the unauthorised sharing of information, or other unwanted forms of agent coordination. Modern steganographic techniques could render such dynamics hard to detect. In this paper, we comprehensively formalise the problem of secret collusion in systems of generative AI agents by drawing on relevant concepts from both the AI and security literature. We study incentives for the use of steganography, and propose a variety of mitigation measures. Our investigations result in a model evaluation framework that systematically tests capabilities required for various forms of secret collusion. We provide extensive empirical results across a range of contemporary LLMs. While the steganographic capabilities of current models remain limited, GPT-4 displays a capability jump suggesting the need for continuous monitoring of steganographic frontier model capabilities. We conclude by laying out a comprehensive research program to mitigate future risks of collusion between generative AI models.

The increasing prevalence of multi-agent learning systems in society necessitates understanding how to learn effective and safe policies in general-sum multi-agent environments against a variety of opponents, including self-play. General-sum learning is difficult because of non-stationary opponents and misaligned incentives. Our first main contribution is to show that many recent approaches to general-sum learning can be derived as approximations to Stackelberg strategies, which suggests a framework for developing new multi-agent learning algorithms. We then define non-coincidental games as games in which the Stackelberg strategy profile is not a Nash Equilibrium. This notably includes several canonical matrix games and provides a normative theory for why existing algorithms fail in self-play in such games. We address this problem by introducing Welfare Equilibria (WE) as a generalisation of Stackelberg Strategies, which can recover desirable Nash Equilibria even in non-coincidental games. Finally, we introduce Welfare Function Search (WelFuSe) as a practical approach to finding desirable WE against unknown opponents, which finds more mutually desirable solutions in self-play, while preserving performance against naive learning opponents., Comment: 31 pages, 23 figures

The rapid proliferation of open-source language models significantly increases the risks of downstream backdoor attacks. These backdoors can introduce dangerous behaviours during model deployment and can evade detection by conventional cybersecurity monitoring systems. In this paper, we introduce a novel class of backdoors in autoregressive transformer models, that, in contrast to prior art, are unelicitable in nature. Unelicitability prevents the defender from triggering the backdoor, making it impossible to evaluate or detect ahead of deployment even if given full white-box access and using automated techniques, such as red-teaming or certain formal verification methods. We show that our novel construction is not only unelicitable thanks to using cryptographic techniques, but also has favourable robustness properties. We confirm these properties in empirical investigations, and provide evidence that our backdoors can withstand state-of-the-art mitigation strategies. Additionally, we expand on previous work by showing that our universal backdoors, while not completely undetectable in white-box settings, can be harder to detect than some existing designs. By demonstrating the feasibility of seamlessly integrating backdoors into transformer models, this paper fundamentally questions the efficacy of pre-deployment detection strategies. This offers new insights into the offence-defence balance in AI safety and security., Comment: 10 pages, 5 figures

Minimum-entropy coupling (MEC) -- the process of finding a joint distribution with minimum entropy for given marginals -- has applications in areas such as causality and steganography. However, existing algorithms are either computationally intractable for large-support distributions or limited to specific distribution types and sensitive to hyperparameter choices. This work addresses these limitations by unifying a prior family of iterative MEC (IMEC) approaches into a generalized partition-based formalism. From this framework, we derive a novel IMEC algorithm called ARIMEC, capable of handling arbitrary discrete distributions, and introduce a method to make IMEC robust to suboptimal hyperparameter settings. These innovations facilitate the application of IMEC to high-throughput steganography with language models, among other settings. Our codebase is available at https://github.com/ssokota/mec .

While reinforcement learning (RL) algorithms have been successfully applied across numerous sequential decision-making problems, their generalization to unforeseen testing environments remains a significant concern. In this paper, we study the problem of out-of-distribution (OOD) detection in RL, which focuses on identifying situations at test time that RL agents have not encountered in their training environments. We first propose a clarification of terminology for OOD detection in RL, which aligns it with the literature from other machine learning domains. We then present new benchmark scenarios for OOD detection, which introduce anomalies with temporal autocorrelation into different components of the agent-environment loop. We argue that such scenarios have been understudied in the current literature, despite their relevance to real-world situations. Confirming our theoretical predictions, our experimental results suggest that state-of-the-art OOD detectors are not able to identify such anomalies. To address this problem, we propose a novel method for OOD detection, which we call DEXTER (Detection via Extraction of Time Series Representations). By treating environment observations as time series data, DEXTER extracts salient time series features, and then leverages an ensemble of isolation forest algorithms to detect anomalies. We find that DEXTER can reliably identify anomalies across benchmark scenarios, exhibiting superior performance compared to both state-of-the-art OOD detectors and high-dimensional changepoint detectors adopted from statistics., Comment: Accepted as a full paper to the 23rd International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2024)

In the next few years, applications of Generative AI are expected to revolutionize a number of different areas, ranging from science & medicine to education. The potential for these seismic changes has triggered a lively debate about potential risks and resulted in calls for tighter regulation, in particular from some of the major tech companies who are leading in AI development. This regulation is likely to put at risk the budding field of open-source Generative AI. We argue for the responsible open sourcing of generative AI models in the near and medium term. To set the stage, we first introduce an AI openness taxonomy system and apply it to 40 current large language models. We then outline differential benefits and risks of open versus closed source AI and present potential risk mitigation, ranging from best practices to calls for technical and scientific contributions. We hope that this report will add a much needed missing voice to the current public discourse on near to mid-term AI safety and other societal impact., Comment: Accepted to ICML'24 as a position paper

By enabling agents to communicate, recent cooperative multi-agent reinforcement learning (MARL) methods have demonstrated better task performance and more coordinated behavior. Most existing approaches facilitate inter-agent communication by allowing agents to send messages to each other through free communication channels, i.e., cheap talk channels. Current methods require these channels to be constantly accessible and known to the agents a priori. In this work, we lift these requirements such that the agents must discover the cheap talk channels and learn how to use them. Hence, the problem has two main parts: cheap talk discovery (CTD) and cheap talk utilization (CTU). We introduce a novel conceptual framework for both parts and develop a new algorithm based on mutual information maximization that outperforms existing algorithms in CTD/CTU settings. We also release a novel benchmark suite to stimulate future research in CTD/CTU., Comment: The 11th International Conference on Learning Representations (ICLR)

Benchmarks play an important role in the development of machine learning algorithms. For example, research in reinforcement learning (RL) has been heavily influenced by available environments and benchmarks. However, RL environments are traditionally run on the CPU, limiting their scalability with typical academic compute. Recent advancements in JAX have enabled the wider use of hardware acceleration to overcome these computational hurdles, enabling massively parallel RL training pipelines and environments. This is particularly useful for multi-agent reinforcement learning (MARL) research. First of all, multiple agents must be considered at each environment step, adding computational burden, and secondly, the sample complexity is increased due to non-stationarity, decentralised partial observability, or other MARL challenges. In this paper, we present JaxMARL, the first open-source code base that combines ease-of-use with GPU enabled efficiency, and supports a large number of commonly used MARL environments as well as popular baseline algorithms. When considering wall clock time, our experiments show that per-run our JAX-based training pipeline is up to 12500x faster than existing approaches. This enables efficient and thorough evaluations, with the potential to alleviate the evaluation crisis of the field. We also introduce and benchmark SMAX, a vectorised, simplified version of the popular StarCraft Multi-Agent Challenge, which removes the need to run the StarCraft II game engine. This not only enables GPU acceleration, but also provides a more flexible MARL environment, unlocking the potential for self-play, meta-learning, and other future applications in MARL. We provide code at https://github.com/flairox/jaxmarl.

Bayesian reinforcement learning (RL) offers a principled and elegant approach for sequential decision making under uncertainty. Most notably, Bayesian agents do not face an exploration/exploitation dilemma, a major pathology of frequentist methods. However theoretical understanding of model-free approaches is lacking. In this paper, we introduce a novel Bayesian model-free formulation and the first analysis showing that model-free approaches can yield Bayes-optimal policies. We show all existing model-free approaches make approximations that yield policies that can be arbitrarily Bayes-suboptimal. As a first step towards model-free Bayes optimality, we introduce the Bayesian exploration network (BEN) which uses normalising flows to model both the aleatoric uncertainty (via density estimation) and epistemic uncertainty (via variational inference) in the Bellman operator. In the limit of complete optimisation, BEN learns true Bayes-optimal policies, but like in variational expectation-maximisation, partial optimisation renders our approach tractable. Empirical results demonstrate that BEN can learn true Bayes-optimal policies in tasks where existing model-free approaches fail., Comment: ICML 2024 Version Update

Der Band geht exemplarisch und interdisziplinär den medial und diskursiv artikulierten Zusammenhängen zwischen Selbst- und Fremdwahrnehmung sowie individuell-subjektivem Bekennen in ihren frühneuzeitlichen Ausprägungen nach. Er beleuchtet grundsätzlich das Zusammenspiel von Bekenntnisakt und Wahrnehmungsformation und untersucht es auf seine Realisations- und Artikulationsformen in der Frühen Neuzeit hin. Angelegt ist er als Studienbuch, das an ausgewählten Quellenbeispielen interdisziplinär entwickelte heuristische Instrumentarien zur Anwendung bringt und so erprobt. Zu diesem Zweck werden inter- und innerkonfessionelle Diskurse zwischen Persönlichkeiten und Institutionen in den Blick genommen, wie sie in theologischen Traktaten, der Geschichtsschreibung, Flugblättern, Vertragstexten oder in der Darstellung von Festinszenierungen ihren Ausdruck finden. Die durch den Band geleistete Diskussion und Zusammenführung von Fallbeispielen der frühneuzeitlichen europäischen Geschichte zielt dabei auch auf die Anwendung des wahrnehmungshistorischen Ansatzes im Kontext akademischer Lehre ab.

The energy transition potentially poses an existential risk for major international oil companies (IOCs) if they fail to adapt to low-carbon business models. Projections of energy futures, however, are met with diverging assumptions on its scale and pace, causing disagreement among IOC decision-makers and their stakeholders over what the business model of an incumbent fossil fuel company should be. In this work, we used deep multi-agent reinforcement learning to solve an energy systems wargame wherein players simulate IOC decision-making, including hydrocarbon and low-carbon investments decisions, dividend policies, and capital structure measures, through an uncertain energy transition to explore critical and non-linear governance questions, from leveraged transitions to reserve replacements. Adversarial play facilitated by state-of-the-art algorithms revealed decision-making strategies robust to energy transition uncertainty and against multiple IOCs. In all games, robust strategies emerged in the form of low-carbon business models as a result of early transition-oriented movement. IOCs adopting such strategies outperformed business-as-usual and delayed transition strategies regardless of hydrocarbon demand projections. In addition to maximizing value, these strategies benefit greater society by contributing substantial amounts of capital necessary to accelerate the global low-carbon energy transition. Our findings point towards the need for lenders and investors to effectively mobilize transition-oriented finance and engage with IOCs to ensure responsible reallocation of capital towards low-carbon business models that would enable the emergence of fossil fuel incumbents as future low-carbon leaders.

Steganography is the practice of encoding secret information into innocuous content in such a manner that an adversarial third party would not realize that there is hidden meaning. While this problem has classically been studied in security literature, recent advances in generative models have led to a shared interest among security and machine learning researchers in developing scalable steganography techniques. In this work, we show that a steganography procedure is perfectly secure under Cachin (1998)'s information-theoretic model of steganography if and only if it is induced by a coupling. Furthermore, we show that, among perfectly secure procedures, a procedure maximizes information throughput if and only if it is induced by a minimum entropy coupling. These insights yield what are, to the best of our knowledge, the first steganography algorithms to achieve perfect security guarantees for arbitrary covertext distributions. To provide empirical validation, we compare a minimum entropy coupling-based approach to three modern baselines -- arithmetic coding, Meteor, and adaptive dynamic grouping -- using GPT-2, WaveRNN, and Image Transformer as communication channels. We find that the minimum entropy coupling-based approach achieves superior encoding efficiency, despite its stronger security constraints. In aggregate, these results suggest that it may be natural to view information-theoretic steganography through the lens of minimum entropy coupling.

Successful coordination in Dec-POMDPs requires agents to adopt robust strategies and interpretable styles of play for their partner. A common failure mode is symmetry breaking, when agents arbitrarily converge on one out of many equivalent but mutually incompatible policies. Commonly these examples include partial observability, e.g. waving your right hand vs. left hand to convey a covert message. In this paper, we present a novel equivariant network architecture for use in Dec-POMDPs that effectively leverages environmental symmetry for improving zero-shot coordination, doing so more effectively than prior methods. Our method also acts as a ``coordination-improvement operator'' for generic, pre-trained policies, and thus may be applied at test-time in conjunction with any self-play algorithm. We provide theoretical guarantees of our work and test on the AI benchmark task of Hanabi, where we demonstrate our methods outperforming other symmetry-aware baselines in zero-shot coordination, as well as able to improve the coordination ability of a variety of pre-trained policies. In particular, we show our method can be used to improve on the state of the art for zero-shot coordination on the Hanabi benchmark.

Tremendous progress has been made in reinforcement learning (RL) over the past decade. Most of these advancements came through the continual development of new algorithms, which were designed using a combination of mathematical derivations, intuitions, and experimentation. Such an approach of creating algorithms manually is limited by human understanding and ingenuity. In contrast, meta-learning provides a toolkit for automatic machine learning method optimisation, potentially addressing this flaw. However, black-box approaches which attempt to discover RL algorithms with minimal prior structure have thus far not outperformed existing hand-crafted algorithms. Mirror Learning, which includes RL algorithms, such as PPO, offers a potential middle-ground starting point: while every method in this framework comes with theoretical guarantees, components that differentiate them are subject to design. In this paper we explore the Mirror Learning space by meta-learning a "drift" function. We refer to the immediate result as Learnt Policy Optimisation (LPO). By analysing LPO we gain original insights into policy optimisation which we use to formulate a novel, closed-form RL algorithm, Discovered Policy Optimisation (DPO). Our experiments in Brax environments confirm state-of-the-art performance of LPO and DPO, as well as their transfer to unseen settings., Comment: NeurIPS 2022

Autonomous agents deployed in the real world need to be robust against adversarial attacks on sensory inputs. Robustifying agent policies requires anticipating the strongest attacks possible. We demonstrate that existing observation-space attacks on reinforcement learning agents have a common weakness: while effective, their lack of information-theoretic detectability constraints makes them detectable using automated means or human inspection. Detectability is undesirable to adversaries as it may trigger security escalations. We introduce {\epsilon}-illusory, a novel form of adversarial attack on sequential decision-makers that is both effective and of {\epsilon}-bounded statistical detectability. We propose a novel dual ascent algorithm to learn such attacks end-to-end. Compared to existing attacks, we empirically find {\epsilon}-illusory to be significantly harder to detect with automated methods, and a small study with human participants (IRB approval under reference R84123/RE001) suggests they are similarly harder to detect for humans. Our findings suggest the need for better anomaly detectors, as well as effective hardware- and system-level defenses. The project website can be found at https://tinyurl.com/illusory-attacks., Comment: ICLR 2024 Spotlight (top 5%)

Surround-view fisheye cameras are commonly used for near-field sensing in automated driving. Four fisheye cameras on four sides of the vehicle are sufficient to cover 360{\deg} around the vehicle capturing the entire near-field region. Some primary use cases are automated parking, traffic jam assist, and urban driving. There are limited datasets and very little work on near-field perception tasks as the focus in automotive perception is on far-field perception. In contrast to far-field, surround-view perception poses additional challenges due to high precision object detection requirements of 10cm and partial visibility of objects. Due to the large radial distortion of fisheye cameras, standard algorithms cannot be extended easily to the surround-view use case. Thus, we are motivated to provide a self-contained reference for automotive fisheye camera perception for researchers and practitioners. Firstly, we provide a unified and taxonomic treatment of commonly used fisheye camera models. Secondly, we discuss various perception tasks and existing literature. Finally, we discuss the challenges and future direction., Comment: Accepted for publication at IEEE Transactions on Intelligent Transportation Systems

In general-sum games, the interaction of self-interested learning agents commonly leads to collectively worst-case outcomes, such as defect-defect in the iterated prisoner's dilemma (IPD). To overcome this, some methods, such as Learning with Opponent-Learning Awareness (LOLA), shape their opponents' learning process. However, these methods are myopic since only a small number of steps can be anticipated, are asymmetric since they treat other agents as naive learners, and require the use of higher-order derivatives, which are calculated through white-box access to an opponent's differentiable learning algorithm. To address these issues, we propose Model-Free Opponent Shaping (M-FOS). M-FOS learns in a meta-game in which each meta-step is an episode of the underlying inner game. The meta-state consists of the inner policies, and the meta-policy produces a new inner policy to be used in the next episode. M-FOS then uses generic model-free optimisation methods to learn meta-policies that accomplish long-horizon opponent shaping. Empirically, M-FOS near-optimally exploits naive learners and other, more sophisticated algorithms from the literature. For example, to the best of our knowledge, it is the first method to learn the well-known Zero-Determinant (ZD) extortion strategy in the IPD. In the same settings, M-FOS leads to socially optimal outcomes under meta-self-play. Finally, we show that M-FOS can be scaled to high-dimensional settings., Comment: ICML 2022 camera ready version. Code: https://github.com/luchris429/Model-Free-Opponent-Shaping

Insufficient Social Cost of Carbon (SCC) estimation methods and short-term decision-making horizons have hindered the ability of carbon emitters to properly correct for the negative externalities of climate change, as well as the capacity of nations to balance economic and climate policy. To overcome these limitations, we introduce Retrospective Social Cost of Carbon Updating (ReSCCU), a novel mechanism that corrects for these limitations as empirically measured evidence is collected. To implement ReSCCU in the context of carbon taxation, we propose Retroactive Carbon Pricing (ReCaP), a market mechanism in which polluters offload the payment of ReSCCU adjustments to insurers. To alleviate systematic risks and minimize government involvement, we introduce the Private ReCaP (PReCaP) prediction market, which could see real-world implementation based on the engagement of a few high net-worth individuals or independent institutions.

Modern deep reinforcement learning (RL) algorithms are motivated by either the generalised policy iteration (GPI) or trust-region learning (TRL) frameworks. However, algorithms that strictly respect these theoretical frameworks have proven unscalable. Surprisingly, the only known scalable algorithms violate the GPI/TRL assumptions, e.g. due to required regularisation or other heuristics. The current explanation of their empirical success is essentially "by analogy": they are deemed approximate adaptations of theoretically sound methods. Unfortunately, studies have shown that in practice these algorithms differ greatly from their conceptual ancestors. In contrast, in this paper we introduce a novel theoretical framework, named Mirror Learning, which provides theoretical guarantees to a large class of algorithms, including TRPO and PPO. While the latter two exploit the flexibility of our framework, GPI and TRL fit in merely as pathologically restrictive corner cases thereof. This suggests that the empirical performance of state-of-the-art methods is a direct consequence of their theoretical properties, rather than of aforementioned approximate analogies. Mirror learning sets us free to boldly explore novel, theoretically sound RL algorithms, a thus far uncharted wonderland.

Self-play is a common paradigm for constructing solutions in Markov games that can yield optimal policies in collaborative settings. However, these policies often adopt highly-specialized conventions that make playing with a novel partner difficult. To address this, recent approaches rely on encoding symmetry and convention-awareness into policy training, but these require strong environmental assumptions and can complicate policy training. We therefore propose moving the learning of conventions to the belief space. Specifically, we propose a belief learning model that can maintain beliefs over rollouts of policies not seen at training time, and can thus decode and adapt to novel conventions at test time. We show how to leverage this model for both search and training of a best response over various pools of policies to greatly improve ad-hoc teamplay. We also show how our setup promotes explainability and interpretability of nuanced agent conventions.

A physically-motivated genetic algorithm (GA) and full enumeration for a tile-based model of self-assembly (JaTAM) is implemented using a graphics processing unit (GPU). We observe performance gains with respect to state-of-the-art implementations on CPU of factor 7.7 for the GA and 2.9 for JaTAM. The correctness of our GA implementation is demonstrated using a test-bed fitness function, and our JaTAM implementation is verified by classifying a well-known search space $S_{2,8}$ based on two tile types. The performance gains achieved allow for the classification of a larger search space $S^{32}_{3,8}$ based on three tile types. The prevalence of structures based on two tile types demonstrates that simple organisms emerge preferrably even in complex ecosystems. The modularity of the largest structures found motivates the assumption that to first order, $S_{2,8}$ forms the building blocks of $S_{3,8}$. We conclude that GPUs may play an important role in future studies of evolutionary dynamics., Comment: MPhys Thesis, 2012. Awarded University of Oxford Tessella Prize

Real world multi-agent tasks often involve varying types and quantities of agents and non-agent entities; however, agents within these tasks rarely need to consider all others at all times in order to act effectively. Factored value function approaches have historically leveraged such independences to improve learning efficiency, but these approaches typically rely on domain knowledge to select fixed subsets of state features to include in each factor. We propose to utilize value function factoring with random subsets of entities in each factor as an auxiliary objective in order to disentangle value predictions from irrelevant entities. This factoring approach is instantiated through a simple attention mechanism masking procedure. We hypothesize that such an approach helps agents learn more effectively in multi-agent settings by discovering common trajectories across episodes within sub-groups of agents/entities. Our approach, Randomized Entity-wise Factorization for Imagined Learning (REFIL), outperforms all strong baselines by a significant margin in challenging StarCraft micromanagement tasks., Algorithmics

Airborne transmission by droplets and aerosols is important for the spread of viruses. Face masks are a well-established preventive measure, but their effectiveness for mitigating SARS-CoV-2 transmission is still under debate. We show that variations in mask efficacy can be explained by different regimes of virus abundance and related to population-average infection probability and reproduction number. For SARS-CoV-2, the viral load of infectious individuals can vary by orders of magnitude. We find that most environments and contacts are under conditions of low virus abundance (virus-limited) where surgical masks are effective at preventing virus spread. More advanced masks and other protective equipment are required in potentially virus-rich indoor environments including medical centers and hospitals. Masks are particularly effective in combination with other preventive measures like ventilation and distancing.

Cyber attacks are increasing in volume, frequency, and complexity. In response, the security community is looking toward fully automating cyber defense systems using machine learning. However, so far the resultant effects on the coevolutionary dynamics of attackers and defenders have not been examined. In this whitepaper, we hypothesise that increased automation on both sides will accelerate the coevolutionary cycle, thus begging the question of whether there are any resultant fixed points, and how they are characterised. Working within the threat model of Locked Shields, Europe's largest cyberdefense exercise, we study blackbox adversarial attacks on network classifiers. Given already existing attack capabilities, we question the utility of optimal evasion attack frameworks based on minimal evasion distances. Instead, we suggest a novel reinforcement learning setting that can be used to efficiently generate arbitrary adversarial perturbations. We then argue that attacker-defender fixed points are themselves general-sum games with complex phase transitions, and introduce a temporally extended multi-agent reinforcement learning framework in which the resultant dynamics can be studied. We hypothesise that one plausible fixed point of AI-NIDS may be a scenario where the defense strategy relies heavily on whitelisted feature flow subspaces. Finally, we demonstrate that a continual learning approach is required to study attacker-defender dynamics in temporally extended general-sum games.

Real world multi-agent tasks often involve varying types and quantities of agents and non-agent entities; however, agents within these tasks rarely need to consider all others at all times in order to act effectively. Factored value function approaches have historically leveraged such independences to improve learning efficiency, but these approaches typically rely on domain knowledge to select fixed subsets of state features to include in each factor. We propose to utilize value function factoring with random subsets of entities in each factor as an auxiliary objective in order to disentangle value predictions from irrelevant entities. This factoring approach is instantiated through a simple attention mechanism masking procedure. We hypothesize that such an approach helps agents learn more effectively in multi-agent settings by discovering common trajectories across episodes within sub-groups of agents/entities. Our approach, Randomized Entity-wise Factorization for Imagined Learning (REFIL), outperforms all strong baselines by a significant margin in challenging StarCraft micromanagement tasks., Algorithmics

We consider the problem of communicating exogenous information by means of Markov decision process trajectories. This setting, which we call a Markov coding game (MCG), generalizes both source coding and a large class of referential games. MCGs also isolate a problem that is important in decentralized control settings in which cheap-talk is not available -- namely, they require balancing communication with the associated cost of communicating. We contribute a theoretically grounded approach to MCGs based on maximum entropy reinforcement learning and minimum entropy coupling that we call MEME. Due to recent breakthroughs in approximation algorithms for minimum entropy coupling, MEME is not merely a theoretical algorithm, but can be applied to practical settings. Empirically, we show both that MEME is able to outperform a strong baseline on small MCGs and that MEME is able to achieve strong performance on extremely large MCGs. To the latter point, we demonstrate that MEME is able to losslessly communicate binary images via trajectories of Cartpole and Pong, while simultaneously achieving the maximal or near maximal expected returns, and that it is even capable of performing well in the presence of actuator noise., Comment: ICML 2022

In this work we explore an auxiliary loss useful for reinforcement learning in environments where strong performing agents are required to be able to navigate a spatial environment. The auxiliary loss proposed is to minimize the classification error of a neural network classifier that predicts whether or not a pair of states sampled from the agents current episode trajectory are in order. The classifier takes as input a pair of states as well as the agent's memory. The motivation for this auxiliary loss is that there is a strong correlation with which of a pair of states is more recent in the agents episode trajectory and which of the two states is spatially closer to the agent. Our hypothesis is that learning features to answer this question encourages the agent to learn and internalize in memory representations of states that facilitate spatial reasoning. We tested this auxiliary loss on a navigation task in a gridworld and achieved 9.6% increase in accumulative episode reward compared to a strong baseline approach.

Surround View fisheye cameras are commonly deployed in automated driving for 360\deg{} near-field sensing around the vehicle. This work presents a multi-task visual perception network on unrectified fisheye images to enable the vehicle to sense its surrounding environment. It consists of six primary tasks necessary for an autonomous driving system: depth estimation, visual odometry, semantic segmentation, motion segmentation, object detection, and lens soiling detection. We demonstrate that the jointly trained model performs better than the respective single task versions. Our multi-task model has a shared encoder providing a significant computational advantage and has synergized decoders where tasks support each other. We propose a novel camera geometry based adaptation mechanism to encode the fisheye distortion model both at training and inference. This was crucial to enable training on the WoodScape dataset, comprised of data from different parts of the world collected by 12 different cameras mounted on three different cars with different intrinsics and viewpoints. Given that bounding boxes is not a good representation for distorted fisheye images, we also extend object detection to use a polygon with non-uniformly sampled vertices. We additionally evaluate our model on standard automotive datasets, namely KITTI and Cityscapes. We obtain the state-of-the-art results on KITTI for depth estimation and pose estimation tasks and competitive performance on the other tasks. We perform extensive ablation studies on various architecture choices and task weighting methodologies. A short video at https://youtu.be/xbSjZ5OfPes provides qualitative results., Comment: Best Robot Vision paper award finalist (top 4). Camera ready version accepted for RA-L and ICRA 2021 publication

The aim of this study was to investigate whether there is an impact of donation rates on the quality of lungs used for transplantation and whether donor lung quality affects post-transplant outcome in the current Lung Allocation Score era. All consecutive adult LTx performed in Eurotransplant (ET) between January 2012 and December 2016 were included (N = 3053). Donors used for LTx in countries with high donation rate were younger (42% vs. 33% ≤45 years, P < 0.0001), were less often smokers (35% vs. 46%, P < 0.0001), had more often clear chest X-rays (82% vs. 72%, P < 0.0001), had better donor oxygenation ratios (20% vs. 26% with PaO2 /FiO2  ≤ 300 mmHg, P < 0.0001), and had better lung donor score values (LDS; 28% vs. 17% with LDS = 6, P < 0.0001) compared with donors used for LTx in countries with low donation rate. Survival rates for the groups LDS = 6 and ≥7 at 5 years were 69.7% and 60.9% (P = 0.007). Lung donor quality significantly impacts on long-term patient survival. Countries with a low donation rate are more oriented to using donor lungs with a lesser quality compared to countries with a high donation rate. Instead of further stretching donor eligibility criteria, the full potential of the donor pool should be realized.

The aim of this study was to investigate whether there is an impact of donation rates on the quality of lungs used for transplantation and whether donor lung quality affects post-transplant outcome in the current Lung Allocation Score era. All consecutive adult LTx performed in Eurotransplant (ET) between January 2012 and December 2016 were included (N = 3053). Donors used for LTx in countries with high donation rate were younger (42% vs. 33% ≤45 years, P < 0.0001), were less often smokers (35% vs. 46%, P < 0.0001), had more often clear chest X-rays (82% vs. 72%, P < 0.0001), had better donor oxygenation ratios (20% vs. 26% with PaO2 /FiO2  ≤ 300 mmHg, P < 0.0001), and had better lung donor score values (LDS; 28% vs. 17% with LDS = 6, P < 0.0001) compared with donors used for LTx in countries with low donation rate. Survival rates for the groups LDS = 6 and ≥7 at 5 years were 69.7% and 60.9% (P = 0.007). Lung donor quality significantly impacts on long-term patient survival. Countries with a low donation rate are more oriented to using donor lungs with a lesser quality compared to countries with a high donation rate. Instead of further stretching donor eligibility criteria, the full potential of the donor pool should be realized.

In classical computer vision, rectification is an integral part of multi-view depth estimation. It typically includes epipolar rectification and lens distortion correction. This process simplifies the depth estimation significantly, and thus it has been adopted in CNN approaches. However, rectification has several side effects, including a reduced field of view (FOV), resampling distortion, and sensitivity to calibration errors. The effects are particularly pronounced in case of significant distortion (e.g., wide-angle fisheye cameras). In this paper, we propose a generic scale-aware self-supervised pipeline for estimating depth, euclidean distance, and visual odometry from unrectified monocular videos. We demonstrate a similar level of precision on the unrectified KITTI dataset with barrel distortion comparable to the rectified KITTI dataset. The intuition being that the rectification step can be implicitly absorbed within the CNN model, which learns the distortion model without increasing complexity. Our approach does not suffer from a reduced field of view and avoids computational costs for rectification at inference time. To further illustrate the general applicability of the proposed framework, we apply it to wide-angle fisheye cameras with 190$^\circ$ horizontal field of view. The training framework UnRectDepthNet takes in the camera distortion model as an argument and adapts projection and unprojection functions accordingly. The proposed algorithm is evaluated further on the KITTI rectified dataset, and we achieve state-of-the-art results that improve upon our previous work FisheyeDistanceNet. Qualitative results on a distorted test scene video sequence indicate excellent performance https://youtu.be/K6pbx3bU4Ss., Comment: Minor fixes added after IROS 2020 Camera ready submission. IROS 2020 presentation video - https://www.youtube.com/watch?v=3Br2KSWZRrY

Multi-agent settings in the real world often involve tasks with varying types and quantities of agents and non-agent entities; however, common patterns of behavior often emerge among these agents/entities. Our method aims to leverage these commonalities by asking the question: ``What is the expected utility of each agent when only considering a randomly selected sub-group of its observed entities?'' By posing this counterfactual question, we can recognize state-action trajectories within sub-groups of entities that we may have encountered in another task and use what we learned in that task to inform our prediction in the current one. We then reconstruct a prediction of the full returns as a combination of factors considering these disjoint groups of entities and train this ``randomly factorized" value function as an auxiliary objective for value-based multi-agent reinforcement learning. By doing so, our model can recognize and leverage similarities across tasks to improve learning efficiency in a multi-task setting. Our approach, Randomized Entity-wise Factorization for Imagined Learning (REFIL), outperforms all strong baselines by a significant margin in challenging multi-task StarCraft micromanagement settings., Comment: ICML 2021 Camera Ready

The COVID-19 pandemic has highlighted the importance of in-silico epidemiological modelling in predicting the dynamics of infectious diseases to inform health policy and decision makers about suitable prevention and containment strategies. Work in this setting involves solving challenging inference and control problems in individual-based models of ever increasing complexity. Here we discuss recent breakthroughs in machine learning, specifically in simulation-based inference, and explore its potential as a novel venue for model calibration to support the design and evaluation of public health interventions. To further stimulate research, we are developing software interfaces that turn two cornerstone COVID-19 and malaria epidemiology models COVID-sim, (https://github.com/mrc-ide/covid-sim/) and OpenMalaria (https://github.com/SwissTPH/openmalaria) into probabilistic programs, enabling efficient interpretable Bayesian inference within those simulators.

In many real-world settings, a team of agents must coordinate its behaviour while acting in a decentralised fashion. At the same time, it is often possible to train the agents in a centralised fashion where global state information is available and communication constraints are lifted. Learning joint action-values conditioned on extra state information is an attractive way to exploit centralised learning, but the best strategy for then extracting decentralised policies is unclear. Our solution is QMIX, a novel value-based method that can train decentralised policies in a centralised end-to-end fashion. QMIX employs a mixing network that estimates joint action-values as a monotonic combination of per-agent values. We structurally enforce that the joint-action value is monotonic in the per-agent values, through the use of non-negative weights in the mixing network, which guarantees consistency between the centralised and decentralised policies. To evaluate the performance of QMIX, we propose the StarCraft Multi-Agent Challenge (SMAC) as a new benchmark for deep multi-agent reinforcement learning. We evaluate QMIX on a challenging set of SMAC scenarios and show that it significantly outperforms existing multi-agent reinforcement learning methods., Comment: Extended version of the ICML 2018 conference paper (arXiv:1803.11485)

We propose FACtored Multi-Agent Centralised policy gradients (FACMAC), a new method for cooperative multi-agent reinforcement learning in both discrete and continuous action spaces. Like MADDPG, a popular multi-agent actor-critic method, our approach uses deep deterministic policy gradients to learn policies. However, FACMAC learns a centralised but factored critic, which combines per-agent utilities into the joint action-value function via a non-linear monotonic function, as in QMIX, a popular multi-agent Q-learning algorithm. However, unlike QMIX, there are no inherent constraints on factoring the critic. We thus also employ a nonmonotonic factorisation and empirically demonstrate that its increased representational capacity allows it to solve some tasks that cannot be solved with monolithic, or monotonically factored critics. In addition, FACMAC uses a centralised policy gradient estimator that optimises over the entire joint action space, rather than optimising over each agent's action space separately as in MADDPG. This allows for more coordinated policy changes and fully reaps the benefits of a centralised critic. We evaluate FACMAC on variants of the multi-agent particle environments, a novel multi-agent MuJoCo benchmark, and a challenging set of StarCraft II micromanagement tasks. Empirical results demonstrate FACMAC's superior performance over MADDPG and other baselines on all three domains.

Extreme precipitation events, such as violent rainfall and hail storms, routinely ravage economies and livelihoods around the developing world. Climate change further aggravates this issue. Data-driven deep learning approaches could widen the access to accurate multi-day forecasts, to mitigate against such events. However, there is currently no benchmark dataset dedicated to the study of global precipitation forecasts. In this paper, we introduce \textbf{RainBench}, a new multi-modal benchmark dataset for data-driven precipitation forecasting. It includes simulated satellite data, a selection of relevant meteorological data from the ERA5 reanalysis product, and IMERG precipitation data. We also release \textbf{PyRain}, a library to process large precipitation datasets efficiently. We present an extensive analysis of our novel dataset and establish baseline results for two benchmark medium-range precipitation forecasting tasks. Finally, we discuss existing data-driven weather forecasting methodologies and suggest future research avenues., Comment: Work completed during the 2020 Frontier Development Lab research accelerator, a private-public partnership with NASA in the US, and ESA in Europe. Accepted as a spotlight/long oral talk at both Climate Change and AI, as well as AI for Earth Sciences Workshops at NeurIPS 2020

Most recently developed approaches to cooperative multi-agent reinforcement learning in the \emph{centralized training with decentralized execution} setting involve estimating a centralized, joint value function. In this paper, we demonstrate that, despite its various theoretical shortcomings, Independent PPO (IPPO), a form of independent learning in which each agent simply estimates its local value function, can perform just as well as or better than state-of-the-art joint learning approaches on popular multi-agent benchmark suite SMAC with little hyperparameter tuning. We also compare IPPO to several variants; the results suggest that IPPO's strong performance may be due to its robustness to some forms of environment non-stationarity.

PURPOSE: Following introduction of the lung allocation score (LAS) in 2011, Eurotransplant member centers can apply for an exceptional LAS (eLAS) if the calculated LAS insufficiently reflects the perceived transplant benefit for a patient, specifically in case of primary pulmonary hypertension group 1 and 4; combined lung+non-renal transplantation; rare diseases; or extracorporeal support. Each eLAS proposal is evaluated by a LAS Review Board, consisting of ≥3 lung transplant experts, which subsequently declines or approves the eLAS request in consensus of ≥3 votes. In case of a lower than accepted score, predefined business rules to assign LAS percentiles are used. METHODS: A retrospective analysis of all eLAS requests in Eurotransplant from December 2011 until September 2019. RESULTS: Overall, 5183 lung transplants (deceased donors) were performed and 420 eLAS requests were made (Germany 52%, Netherlands 18%, Austria 18%, Belgium 13%), of which 116 (28%) were approved. Most eLAS requests concerned group B/Pulmonary vascular disease (44%), followed by group C/Cystic fibrosis or immunodeficiency disorder (28%), then group D/Restrictive lung disease (15%) and finally group A/Obstructive lung disease (11%); whereas 10 patients (2%) were not classified. The proportion of accepted eLAS requests significantly differed between countries (Germany 25%, Netherlands 37%, Austria 20%, Belgium 36%) (p=0.042). eLAS requests decreased in the Netherlands following its LAS introduction in 2014 (2011-2014 mean 13/yr vs. 2015-2019 mean 4.6/yr; p=0.060). However, since 2015 an overall annual increasing number of eLAS requests is seen, with doubling of the eLAS requests in 2018 vs. 2015, but no difference in acceptance rate (2015-2018: 22.4%) (Figure). Acceptance rates were 38% for Group B, 21% for Group C, 20% for Group D and 11% for Group A. CONCLUSION: The observed variations require further investigation to optimize lung allocation for specific patient populations in Eurotransplant, SCOPUS: ar.j, info:eu-repo/semantics/published

The aim of this study was to investigate whether there is an impact of donation rates on the quality of lungs used for transplantation and whether donor lung quality affects post-transplant outcome in the current Lung Allocation Score era. All consecutive adult LTx performed in Eurotransplant (ET) between January 2012 and December 2016 were included (N = 3053). Donors used for LTx in countries with high donation rate were younger (42% vs. 33% ≤45 years, P < 0.0001), were less often smokers (35% vs. 46%, P < 0.0001), had more often clear chest X-rays (82% vs. 72%, P < 0.0001), had better donor oxygenation ratios (20% vs. 26% with PaO2/FiO2 ≤ 300 mmHg, P < 0.0001), and had better lung donor score values (LDS; 28% vs. 17% with LDS = 6, P < 0.0001) compared with donors used for LTx in countries with low donation rate. Survival rates for the groups LDS = 6 and ≥7 at 5 years were 69.7% and 60.9% (P = 0.007). Lung donor quality significantly impacts on long-term patient survival. Countries with a low donation rate are more oriented to using donor lungs with a lesser quality compared to countries with a high donation rate. Instead of further stretching donor eligibility criteria, the full potential of the donor pool should be realized., SCOPUS: ar.j, info:eu-repo/semantics/published

Naive approaches to amortized inference in probabilistic programs with unbounded loops can produce estimators with infinite variance. This is particularly true of importance sampling inference in programs that explicitly include rejection sampling as part of the user-programmed generative procedure. In this paper we develop a new and efficient amortized importance sampling estimator. We prove finite variance of our estimator and empirically demonstrate our method's correctness and efficiency compared to existing alternatives on generative programs containing rejection sampling loops and discuss how to implement our method in a generic probabilistic programming framework., Comment: AISTATS 2022 camera ready

Fisheye cameras are commonly used in applications like autonomous driving and surveillance to provide a large field of view ($>180^{\circ}$). However, they come at the cost of strong non-linear distortions which require more complex algorithms. In this paper, we explore Euclidean distance estimation on fisheye cameras for automotive scenes. Obtaining accurate and dense depth supervision is difficult in practice, but self-supervised learning approaches show promising results and could potentially overcome the problem. We present a novel self-supervised scale-aware framework for learning Euclidean distance and ego-motion from raw monocular fisheye videos without applying rectification. While it is possible to perform piece-wise linear approximation of fisheye projection surface and apply standard rectilinear models, it has its own set of issues like re-sampling distortion and discontinuities in transition regions. To encourage further research in this area, we will release our dataset as part of the WoodScape project \cite{yogamani2019woodscape}. We further evaluated the proposed algorithm on the KITTI dataset and obtained state-of-the-art results comparable to other self-supervised monocular methods. Qualitative results on an unseen fisheye video demonstrate impressive performance https://youtu.be/Sgq1WzoOmXg., Comment: Minor fixes added after ICRA 2020 camera ready submission. ICRA 2020 presentation video - https://www.youtube.com/watch?v=qAsdpHP5e8c&t

Epidemiology simulations have become a fundamental tool in the fight against the epidemics of various infectious diseases like AIDS and malaria. However, the complicated and stochastic nature of these simulators can mean their output is difficult to interpret, which reduces their usefulness to policymakers. In this paper, we introduce an approach that allows one to treat a large class of population-based epidemiology simulators as probabilistic generative models. This is achieved by hijacking the internal random number generator calls, through the use of a universal probabilistic programming system (PPS). In contrast to other methods, our approach can be easily retrofitted to simulators written in popular industrial programming frameworks. We demonstrate that our method can be used for interpretable introspection and inference, thus shedding light on black-box simulators. This reinstates much-needed trust between policymakers and evidence-based methods., Comment: 6 pages, 3 figures, Accepted at the International Conference on Machine Learning AI for Social Good Workshop, Long Beach, United States, 2019

As global greenhouse gas emissions continue to rise, the use of stratospheric aerosol injection (SAI), a form of solar geoengineering, is increasingly considered in order to artificially mitigate climate change effects. However, initial research in simulation suggests that naive SAI can have catastrophic regional consequences, which may induce serious geostrategic conflicts. Current geo-engineering research treats SAI control in low-dimensional approximation only. We suggest treating SAI as a high-dimensional control problem, with policies trained according to a context-sensitive reward function within the Deep Reinforcement Learning (DRL) paradigm. In order to facilitate training in simulation, we suggest to emulate HadCM3, a widely used General Circulation Model, using deep learning techniques. We believe this is the first application of DRL to the climate sciences., Comment: Awarded Poster and Spotlight Oral at Climate Change: How Can AI Help? (Workshop) at International Conference on Machine Learning, Long Beach, California, 2019

Fisheye cameras are commonly employed for obtaining a large field of view in surveillance, augmented reality and in particular automotive applications. In spite of their prevalence, there are few public datasets for detailed evaluation of computer vision algorithms on fisheye images. We release the first extensive fisheye automotive dataset, WoodScape, named after Robert Wood who invented the fisheye camera in 1906. WoodScape comprises of four surround view cameras and nine tasks including segmentation, depth estimation, 3D bounding box detection and soiling detection. Semantic annotation of 40 classes at the instance level is provided for over 10,000 images and annotation for other tasks are provided for over 100,000 images. With WoodScape, we would like to encourage the community to adapt computer vision models for fisheye camera instead of using naive rectification., Comment: Accepted for Oral Presentation at IEEE International Conference on Computer Vision (ICCV) 2019. Please refer to our website https://woodscape.valeo.com and https://github.com/valeoai/woodscape for release status and updates

In the last few years, deep multi-agent reinforcement learning (RL) has become a highly active area of research. A particularly challenging class of problems in this area is partially observable, cooperative, multi-agent learning, in which teams of agents must learn to coordinate their behaviour while conditioning only on their private observations. This is an attractive research area since such problems are relevant to a large number of real-world systems and are also more amenable to evaluation than general-sum problems. Standardised environments such as the ALE and MuJoCo have allowed single-agent RL to move beyond toy domains, such as grid worlds. However, there is no comparable benchmark for cooperative multi-agent RL. As a result, most papers in this field use one-off toy problems, making it difficult to measure real progress. In this paper, we propose the StarCraft Multi-Agent Challenge (SMAC) as a benchmark problem to fill this gap. SMAC is based on the popular real-time strategy game StarCraft II and focuses on micromanagement challenges where each unit is controlled by an independent agent that must act based on local observations. We offer a diverse set of challenge maps and recommendations for best practices in benchmarking and evaluations. We also open-source a deep multi-agent RL learning framework including state-of-the-art algorithms. We believe that SMAC can provide a standard benchmark environment for years to come. Videos of our best agents for several SMAC scenarios are available at: https://youtu.be/VZ7zmQ_obZ0.

Both Eurotransplant (ET) and the US use the lung allocation score (LAS) to allocate donor lungs. In 2015, the US implemented a new algorithm for calculating the score while ET has fine-tuned the original model using business rules. A comparison of both models in a contemporary patient cohort was performed. The rank positions and the correlation between both scores were calculated for all patients on the active waiting list in ET. On February 6th 2017, 581 patients were actively listed on the lung transplant waiting list. The median LAS values were 32.56 and 32.70 in ET and the US, respectively. The overall correlation coefficient between both scores was 0.71. Forty-three per cent of the patients had a < 2 point change in their LAS. US LAS was more than two points lower for 41% and more than two points higher for 16% of the patients. Median ranks and the 90th percentiles for all diagnosis groups did not differ between both scores. Implementing the 2015 US LAS model would not significantly alter the current waiting list in ET.