As a pivotal component to attaining generalizable solutions in human intelligence, reasoning provides great potential for reinforcement learning (RL) agents' generalization towards varied goals by summarizing part-to-whole arguments and discovering cause-and-effect relations. However, how to discover and represent causalities remains a huge gap that hinders the development of causal RL. In this paper, we augment Goal-Conditioned RL (GCRL) with Causal Graph (CG), a structure built upon the relation between objects and events. We novelly formulate the GCRL problem into variational likelihood maximization with CG as latent variables. To optimize the derived objective, we propose a framework with theoretical performance guarantees that alternates between two steps: using interventional data to estimate the posterior of CG; using CG to learn generalizable models and interpretable policies. Due to the lack of public benchmarks that verify generalization capability under reasoning, we design nine tasks and then empirically show the effectiveness of the proposed method against five baselines on these tasks. Further theoretical analysis shows that our performance improvement is attributed to the virtuous cycle of causal discovery, transition modeling, and policy training, which aligns with the experimental evidence in extensive ablation studies.
Humans can leverage prior experience and learn novel tasks from a handful of demonstrations. In contrast to offline meta-reinforcement learning, which aims to achieve quick adaptation through better algorithm design, we investigate the effect of architecture inductive bias on the few-shot learning capability. We propose a Prompt-based Decision Transformer (Prompt-DT), which leverages the sequential modeling ability of the Transformer architecture and the prompt framework to achieve few-shot adaptation in offline RL. We design the trajectory prompt, which contains segments of the few-shot demonstrations, and encodes task-specific information to guide policy generation. Our experiments in five MuJoCo control benchmarks show that Prompt-DT is a strong few-shot learner without any extra finetuning on unseen target tasks. Prompt-DT outperforms its variants and strong meta offline RL baselines by a large margin with a trajectory prompt containing only a few timesteps. Prompt-DT is also robust to prompt length changes and can generalize to out-of-distribution (OOD) environments.
As shown by recent studies, machine intelligence-enabled systems are vulnerable to test cases resulting from either adversarial manipulation or natural distribution shifts. This has raised great concerns about deploying machine learning algorithms for real-world applications, especially in the safety-critical domains such as autonomous driving (AD). On the other hand, traditional AD testing on naturalistic scenarios requires hundreds of millions of driving miles due to the high dimensionality and rareness of the safety-critical scenarios in the real world. As a result, several approaches for autonomous driving evaluation have been explored, which are usually, however, based on different simulation platforms, types of safety-critical scenarios, scenario generation algorithms, and driving route variations. Thus, despite a large amount of effort in autonomous driving testing, it is still challenging to compare and understand the effectiveness and efficiency of different testing scenario generation algorithms and testing mechanisms under similar conditions. In this paper, we aim to provide the first unified platform SafeBench to integrate different types of safety-critical testing scenarios, scenario generation algorithms, and other variations such as driving routes and environments. Meanwhile, we implement 4 deep reinforcement learning-based AD algorithms with 4 types of input (e.g., bird's-eye view, camera) to perform fair comparisons on SafeBench. We find our generated testing scenarios are indeed more challenging and observe the trade-off between the performance of AD agents under benign and safety-critical testing scenarios. We believe our unified platform SafeBench for large-scale and effective autonomous driving testing will motivate the development of new testing scenario generation and safe AD algorithms. SafeBench is available at https://safebench.github.io.
Safe reinforcement learning (RL) trains a policy to maximize the task reward while satisfying safety constraints. While prior works focus on the performance optimality, we find that the optimal solutions of many safe RL problems are not robust and safe against carefully designed observational perturbations. We formally analyze the unique properties of designing effective state adversarial attackers in the safe RL setting. We show that baseline adversarial attack techniques for standard RL tasks are not always effective for safe RL and proposed two new approaches - one maximizes the cost and the other maximizes the reward. One interesting and counter-intuitive finding is that the maximum reward attack is strong, as it can both induce unsafe behaviors and make the attack stealthy by maintaining the reward. We further propose a more effective adversarial training framework for safe RL and evaluate it via comprehensive experiments. This work sheds light on the inherited connection between observational robustness and safety in RL and provides a pioneer work for future safe RL studies.
Self-driving vehicles have their own intelligence to drive on open roads. However, vehicle managers, e.g., government or industrial companies, still need a way to tell these self-driving vehicles what behaviors are encouraged or forbidden. Unlike human drivers, current self-driving vehicles cannot understand the traffic laws, thus rely on the programmers manually writing the corresponding principles into the driving systems. It would be less efficient and hard to adapt some temporary traffic laws, especially when the vehicles use data-driven decision-making algorithms. Besides, current self-driving vehicle systems rarely take traffic law modification into consideration. This work aims to design a road traffic law adaptive decision-making method. The decision-making algorithm is designed based on reinforcement learning, in which the traffic rules are usually implicitly coded in deep neural networks. The main idea is to supply the adaptability to traffic laws of self-driving vehicles by a law-adaptive backup policy. In this work, the natural language-based traffic laws are first translated into a logical expression by the Linear Temporal Logic method. Then, the system will try to monitor in advance whether the self-driving vehicle may break the traffic laws by designing a long-term RL action space. Finally, a sample-based planning method will re-plan the trajectory when the vehicle may break the traffic rules. The method is validated in a Beijing Winter Olympic Lane scenario and an overtaking case, built in CARLA simulator. The results show that by adopting this method, the self-driving vehicles can comply with new issued or updated traffic laws effectively. This method helps self-driving vehicles governed by digital traffic laws, which is necessary for the wide adoption of autonomous driving.
In this paper, we propose a dynamic cascaded encoder Automatic Speech Recognition (ASR) model, which unifies models for different deployment scenarios. Moreover, the model can significantly reduce model size and power consumption without loss of quality. Namely, with the dynamic cascaded encoder model, we explore three techniques to maximally boost the performance of each model size: 1) Use separate decoders for each sub-model while sharing the encoders; 2) Use funnel-pooling to improve the encoder efficiency; 3) Balance the size of causal and non-causal encoders to improve quality and fit deployment constraints. Overall, the proposed large-medium model has 30% smaller size and reduces power consumption by 33%, compared to the baseline cascaded encoder model. The triple-size model that unifies the large, medium, and small models achieves 37% total size reduction with minimal quality loss, while substantially reducing the engineering efforts of having separate models.
Multimedia summarization with multimodal output can play an essential role in real-world applications, i.e., automatically generating cover images and titles for news articles or providing introductions to online videos. In this work, we propose a multimodal hierarchical multimedia summarization (MHMS) framework by interacting visual and language domains to generate both video and textual summaries. Our MHMS method contains video and textual segmentation and summarization module, respectively. It formulates a cross-domain alignment objective with optimal transport distance which leverages cross-domain interaction to generate the representative keyframe and textual summary. We evaluated MHMS on three recent multimodal datasets and demonstrated the effectiveness of our method in producing high-quality multimodal summaries.
Evaluating the performance of autonomous vehicles (AV) and their complex subsystems to high precision under naturalistic circumstances remains a challenge, especially when failure or dangerous cases are rare. Rarity does not only require an enormous sample size for a naive method to achieve high confidence estimation, but it also causes dangerous underestimation of the true failure rate and it is extremely hard to detect. Meanwhile, the state-of-the-art approach that comes with a correctness guarantee can only compute an upper bound for the failure rate under certain conditions, which could limit its practical uses. In this work, we present Deep Importance Sampling (Deep IS) framework that utilizes a deep neural network to obtain an efficient IS that is on par with the state-of-the-art, capable of reducing the required sample size 43 times smaller than the naive sampling method to achieve 10% relative error and while producing an estimate that is much less conservative. Our high-dimensional experiment estimating the misclassification rate of one of the state-of-the-art traffic sign classifiers further reveals that this efficiency still holds true even when the target is very small, achieving over 600 times efficiency boost. This highlights the potential of Deep IS in providing a precise estimate even against high-dimensional uncertainties.
Heart rate variability (HRV) is a practical and noninvasive measure of autonomic nervous system activity, which plays an essential role in cardiovascular health. However, using HRV to assess physiology status is challenging. Even in clinical settings, HRV is sensitive to acute stressors such as physical activity, mental stress, hydration, alcohol, and sleep. Wearable devices provide convenient HRV measurements, but the irregularity of measurements and uncaptured stressors can bias conventional analytical methods. To better interpret HRV measurements for downstream healthcare applications, we learn a personalized diurnal rhythm as an accurate physiological indicator for each individual. We develop Physiological Multitask-Learning (PhysioMTL) by harnessing Optimal Transport theory within a Multitask-learning (MTL) framework. The proposed method learns an individual-specific predictive model from heterogeneous observations, and enables estimation of an optimal transport map that yields a push forward operation onto the demographic features for each task. Our model outperforms competing MTL methodologies on unobserved predictive tasks for synthetic and two real-world datasets. Specifically, our method provides remarkable prediction results on unseen held-out subjects given only $20\%$ of the subjects in real-world observational studies. Furthermore, our model enables a counterfactual engine that generates the effect of acute stressors and chronic conditions on HRV rhythms.