Search

Evaluating policies using off-policy data is crucial for applying reinforcement learning to real-world problems such as healthcare and autonomous driving. Previous methods for off-policy evaluation (OPE) generally suffer from high variance or irreducible bias, leading to unacceptably high prediction errors. In this work, we introduce STAR, a framework for OPE that encompasses a broad range of estimators -- which include existing OPE methods as special cases -- that achieve lower mean squared prediction errors. STAR leverages state abstraction to distill complex, potentially continuous problems into compact, discrete models which we call abstract reward processes (ARPs). Predictions from ARPs estimated from off-policy data are provably consistent (asymptotically correct). Rather than proposing a specific estimator, we present a new framework for OPE and empirically demonstrate that estimators within STAR outperform existing methods. The best STAR estimator outperforms baselines in all twelve cases studied, and even the median STAR estimator surpasses the baselines in seven out of the twelve cases., Comment: Accepted at the Thirty-eighth Annual Conference on Neural Information Processing Systems (NeurIPS 2024)

Novel reinforcement learning algorithms, or improvements on existing ones, are commonly justified by evaluating their performance on benchmark environments and are compared to an ever-changing set of standard algorithms. However, despite numerous calls for improvements, experimental practices continue to produce misleading or unsupported claims. One reason for the ongoing substandard practices is that conducting rigorous benchmarking experiments requires substantial computational time. This work investigates the sources of increased computation costs in rigorous experiment designs. We show that conducting rigorous performance benchmarks will likely have computational costs that are often prohibitive. As a result, we argue for using an additional experimentation paradigm to overcome the limitations of benchmarking., Comment: 19 pages, 13 figures, The Forty-first International Conference on Machine Learning (ICML 2024)

State-of-the-art large language models (LLMs) have become indispensable tools for various tasks. However, training LLMs to serve as effective assistants for humans requires careful consideration. A promising approach is reinforcement learning from human feedback (RLHF), which leverages human feedback to update the model in accordance with human preferences and mitigate issues like toxicity and hallucinations. Yet, an understanding of RLHF for LLMs is largely entangled with initial design choices that popularized the method and current research focuses on augmenting those choices rather than fundamentally improving the framework. In this paper, we analyze RLHF through the lens of reinforcement learning principles to develop an understanding of its fundamentals, dedicating substantial focus to the core component of RLHF -- the reward model. Our study investigates modeling choices, caveats of function approximation, and their implications on RLHF training algorithms, highlighting the underlying assumptions made about the expressivity of reward. Our analysis improves the understanding of the role of reward models and methods for their training, concurrently revealing limitations of the current methodology. We characterize these limitations, including incorrect generalization, model misspecification, and the sparsity of feedback, along with their impact on the performance of a language model. The discussion and analysis are substantiated by a categorical review of current literature, serving as a reference for researchers and practitioners to understand the challenges of RLHF and build upon existing efforts.

In this paper, we introduce a fresh perspective on the challenges of credit assignment and policy evaluation. First, we delve into the nuances of eligibility traces and explore instances where their updates may result in unexpected credit assignment to preceding states. From this investigation emerges the concept of a novel value function, which we refer to as the \emph{bidirectional value function}. Unlike traditional state value functions, bidirectional value functions account for both future expected returns (rewards anticipated from the current state onward) and past expected returns (cumulative rewards from the episode's start to the present). We derive principled update equations to learn this value function and, through experimentation, demonstrate its efficacy in enhancing the process of policy evaluation. In particular, our results indicate that the proposed learning approach can, in certain challenging contexts, perform policy evaluation more rapidly than TD($\lambda$) -- a method that learns forward value functions, $v^\pi$, \emph{directly}. Overall, our findings present a new perspective on eligibility traces and potential advantages associated with the novel value function it inspires, especially for policy evaluation., Comment: Accepted in The 38th Annual AAAI Conference on Artificial Intelligence

Designing reward functions for efficiently guiding reinforcement learning (RL) agents toward specific behaviors is a complex task. This is challenging since it requires the identification of reward structures that are not sparse and that avoid inadvertently inducing undesirable behaviors. Naively modifying the reward structure to offer denser and more frequent feedback can lead to unintended outcomes and promote behaviors that are not aligned with the designer's intended goal. Although potential-based reward shaping is often suggested as a remedy, we systematically investigate settings where deploying it often significantly impairs performance. To address these issues, we introduce a new framework that uses a bi-level objective to learn \emph{behavior alignment reward functions}. These functions integrate auxiliary rewards reflecting a designer's heuristics and domain knowledge with the environment's primary rewards. Our approach automatically determines the most effective way to blend these types of feedback, thereby enhancing robustness against heuristic reward misspecification. Remarkably, it can also adapt an agent's policy optimization process to mitigate suboptimalities resulting from limitations and biases inherent in the underlying RL algorithms. We evaluate our method's efficacy on a diverse set of tasks, from small-scale experiments to high-dimensional control challenges. We investigate heuristic auxiliary rewards of varying quality -- some of which are beneficial and others detrimental to the learning process. Our results show that our framework offers a robust and principled way to integrate designer-specified heuristics. It not only addresses key shortcomings of existing approaches but also consistently leads to high-performing solutions, even when given misaligned or poorly-specified auxiliary reward functions., Comment: (Spotlight) Thirty-seventh Conference on Neural Information Processing Systems (NeurIPS 2023)

Object recognition is a crucial step in perception systems for autonomous and intelligent vehicles, as evidenced by the numerous research works in the topic. In this paper, object recognition is explored by using multisensory and multimodality approaches, with the intention of reducing the false positive rate (FPR). The reduction of the FPR becomes increasingly important in perception systems since the misclassification of an object can potentially cause accidents. In particular, this work presents a strategy through Bayesian inference to reduce the FPR considering the likelihood function as a cumulative distribution function from Gaussian kernel density estimations, and the prior probabilities as cumulative functions of normalized histograms. The validation of the proposed methodology is performed on the KITTI dataset using deep networks (DenseNet, NasNet, and EfficientNet), and recent 3D point cloud networks (PointNet, and PintNet++), by considering three object-categories (cars, cyclists, pedestrians) and the RGB and LiDAR sensor modalities., Comment: This paper has been submitted to the journal Pattern Recognition Letters

Coagent networks for reinforcement learning (RL) [Thomas and Barto, 2011] provide a powerful and flexible framework for deriving principled learning rules for arbitrary stochastic neural networks. The coagent framework offers an alternative to backpropagation-based deep learning (BDL) that overcomes some of backpropagation's main limitations. For example, coagent networks can compute different parts of the network \emph{asynchronously} (at different rates or at different times), can incorporate non-differentiable components that cannot be used with backpropagation, and can explore at levels higher than their action spaces (that is, they can be designed as hierarchical networks for exploration and/or temporal abstraction). However, the coagent framework is not just an alternative to BDL; the two approaches can be blended: BDL can be combined with coagent learning rules to create architectures with the advantages of both approaches. This work generalizes the coagent theory and learning rules provided by previous works; this generalization provides more flexibility for network architecture design within the coagent framework. This work also studies one of the chief disadvantages of coagent networks: high variance updates for networks that have many coagents and do not use backpropagation. We show that a coagent algorithm with a policy network that does not use backpropagation can scale to a challenging RL domain with a high-dimensional state and action space (the MuJoCo Ant environment), learning reasonable (although not state-of-the-art) policies. These contributions motivate and provide a more general theoretical foundation for future work that studies coagent networks.

Methods for sequential decision-making are often built upon a foundational assumption that the underlying decision process is stationary. This limits the application of such methods because real-world problems are often subject to changes due to external factors (passive non-stationarity), changes induced by interactions with the system itself (active non-stationarity), or both (hybrid non-stationarity). In this work, we take the first steps towards the fundamental challenge of on-policy and off-policy evaluation amidst structured changes due to active, passive, or hybrid non-stationarity. Towards this goal, we make a higher-order stationarity assumption such that non-stationarity results in changes over time, but the way changes happen is fixed. We propose, OPEN, an algorithm that uses a double application of counterfactual reasoning and a novel importance-weighted instrument-variable regression to obtain both a lower bias and a lower variance estimate of the structure in the changes of a policy's past performances. Finally, we show promising results on how OPEN can be used to predict future performances for several domains inspired by real-world applications that exhibit non-stationarity., Comment: Accepted at Thirty-sixth Conference on Neural Information Processing Systems (NeurIPS 2022)

Multi-objective reinforcement learning (MORL) algorithms tackle sequential decision problems where agents may have different preferences over (possibly conflicting) reward functions. Such algorithms often learn a set of policies (each optimized for a particular agent preference) that can later be used to solve problems with novel preferences. We introduce a novel algorithm that uses Generalized Policy Improvement (GPI) to define principled, formally-derived prioritization schemes that improve sample-efficient learning. They implement active-learning strategies by which the agent can (i) identify the most promising preferences/objectives to train on at each moment, to more rapidly solve a given MORL problem; and (ii) identify which previous experiences are most relevant when learning a policy for a particular agent preference, via a novel Dyna-style MORL method. We prove our algorithm is guaranteed to always converge to an optimal solution in a finite number of steps, or an $\epsilon$-optimal solution (for a bounded $\epsilon$) if the agent is limited and can only identify possibly sub-optimal policies. We also prove that our method monotonically improves the quality of its partial solutions while learning. Finally, we introduce a bound that characterizes the maximum utility loss (with respect to the optimal solution) incurred by the partial solutions computed by our method throughout learning. We empirically show that our method outperforms state-of-the-art MORL algorithms in challenging multi-objective tasks, both with discrete and continuous state and action spaces., Comment: Accepted to AAMAS 2023

Momentum resolved 4D-STEM, also called center of mass (CoM) analysis, has been used to measure the long range built-in electric field of a silicon p-n junction. The effect of different STEM modes and the trade-off between spatial resolution and electric field sensitivity are studied. Two acquisition modes are compared: nanobeam and low magnification (LM) modes. A thermal noise free Medipix3 direct electron detector with high speed acquisition has been used to study the influence of low electron beam current and millisecond dwell times on the measured electric field and standard deviation. It is shown that LM conditions can underestimate the electric field values due to a bigger probe size used but provide an improvement of almost one order of magnitude on the signal-to-noise ratio, leading to a detection limit of 0.011MV/cm. It is observed that the CoM results do not vary with acquisition time or electron dose as low as 24 $e^-/A^2$, showing that the electron beam does not influence the built-in electric field and that this method can be robust for studying beam sensitive materials, where a low dose is needed.

A key issue in the development of high-performance semiconductor devices is the ability to properly measure active dopants at the nanometer scale. 4D scanning transmission electron microscopy and off-axis electron holography have opened up the possibility of studying the electrostatic properties of a p-n junction with nm-scale spatial resolution. The complete description of a p-n junction must take into account the precise evolution of the concentration of dopants around the junction, since the sharpness of the dopant transition directly influences the built-in potential and the maximum electric field. Here, a contacted silicon p-n junction is studied through in-situ biased 4D-STEM. Measurements of electric field, built-in voltage, depletion region width and charge density in the space charge region are combined with analytical equations as well as finite-element simulations in order to evaluate the quality of the junction interface. The nominally-symmetric, highly doped ($N_A = N_D = 9\space x \space10^{18} cm^{-3}$) junction presents an electric field and built-in voltage much lower than expected for an abrupt junction. These experimental results are consistent with electron holography data. All measured junction parameters are compatible with the presence of an intermediate region with a graded profile of the dopants at the p-n interface. This hypothesis is also consistent with the evolution of the electric field with bias. These results demonstrate that in-situ biased 4D-STEM enables a better understanding of the electrical properties of semiconductor p-n junctions with nm-scale resolution., Comment: 13 pages, 5 figures

We draw on the latest advancements in the physics community to propose a novel method for discovering the governing non-linear dynamics of physical systems in reinforcement learning (RL). We establish that this method is capable of discovering the underlying dynamics using significantly fewer trajectories (as little as one rollout with $\leq 30$ time steps) than state of the art model learning algorithms. Further, the technique learns a model that is accurate enough to induce near-optimal policies given significantly fewer trajectories than those required by model-free algorithms. It brings the benefits of model-based RL without requiring a model to be developed in advance, for systems that have physics-based dynamics. To establish the validity and applicability of this algorithm, we conduct experiments on four classic control tasks. We found that an optimal policy trained on the discovered dynamics of the underlying system can generalize well. Further, the learned policy performs well when deployed on the actual physical system, thus bridging the model to real system gap. We further compare our method to state-of-the-art model-based and model-free approaches, and show that our method requires fewer trajectories sampled on the true physical system compared other methods. Additionally, we explored approximate dynamics models and found that they also can perform well., Comment: 8 pages, 1 figure, 1 table, 1 algorithm, presented at the Decision Awareness in Reinforcement Learning workshop held at the International Conference on Machine Learning, 22 July 2022, Baltimore MD, USA

Recent research has shown that seemingly fair machine learning models, when used to inform decisions that have an impact on peoples' lives or well-being (e.g., applications involving education, employment, and lending), can inadvertently increase social inequality in the long term. This is because prior fairness-aware algorithms only consider static fairness constraints, such as equal opportunity or demographic parity. However, enforcing constraints of this type may result in models that have negative long-term impact on disadvantaged individuals and communities. We introduce ELF (Enforcing Long-term Fairness), the first classification algorithm that provides high-confidence fairness guarantees in terms of long-term, or delayed, impact. We prove that the probability that ELF returns an unfair solution is less than a user-specified tolerance and that (under mild assumptions), given sufficient training data, ELF is able to find and return a fair solution if one exists. We show experimentally that our algorithm can successfully mitigate long-term unfairness., Comment: 24 pages, 5 figures

In many real-world applications, reinforcement learning (RL) agents might have to solve multiple tasks, each one typically modeled via a reward function. If reward functions are expressed linearly, and the agent has previously learned a set of policies for different tasks, successor features (SFs) can be exploited to combine such policies and identify reasonable solutions for new problems. However, the identified solutions are not guaranteed to be optimal. We introduce a novel algorithm that addresses this limitation. It allows RL agents to combine existing policies and directly identify optimal policies for arbitrary new problems, without requiring any further interactions with the environment. We first show (under mild assumptions) that the transfer learning problem tackled by SFs is equivalent to the problem of learning to optimize multiple objectives in RL. We then introduce an SF-based extension of the Optimistic Linear Support algorithm to learn a set of policies whose SFs form a convex coverage set. We prove that policies in this set can be combined via generalized policy improvement to construct optimal behaviors for any new linearly-expressible tasks, without requiring any additional training samples. We empirically show that our method outperforms state-of-the-art competing algorithms both in discrete and continuous domains under value function approximation., Comment: Proceedings of the 39th International Conference on Machine Learning (ICML'22)