Driving in dense traffic with human and autonomous drivers is a challenging task that requires high-level planning and reasoning. Human drivers can achieve this task comfortably, and there has been many efforts to model human driver strategies. These strategies can be used as inspirations for developing autonomous driving algorithms or to create high-fidelity simulators. Reinforcement learning is a common tool to model driver policies, but conventional training of these models can be computationally expensive and time-consuming. To address this issue, in this paper, we propose ``skill-based" hierarchical driving strategies, where motion primitives, i.e. skills, are designed and used as high-level actions. This reduces the training time for applications that require multiple models with varying behavior. Simulation results in a merging scenario demonstrate that the proposed approach yields driver models that achieve higher performance with less training compared to baseline reinforcement learning methods.

Acknowledgments: The authors would like to acknowledge primary support from the National Science Foundation National Robotics Initiative which has made this research possible, additional support from NASA's PSTAR program, and in-kind support by the NOAA Ocean Exploration Cooperative Institute with ship and robotic vehicle operations during 2021 Pacific Ocean demonstrations in the San Pedro Basin. The authors would also like to thank the captain and crew of the R/V Nautilus, the NUI robotic vehicle operations team, and study participants who volunteered to assist with performance testing of the SHARC and conventional robotic manipulation systems. AP would like to acknowledge support from the National Science Foundation Graduate Research Fellowship under Grant No. 2141064 and the Link Foundation. Funding: National Science Foundation, National Robotics Initiative grant IIS-1830500 (RC) National Science Foundation, National Robotics Initiative grant IIS-1830660 (MW) National Aeronautics and Space Administration, Planetary Science and Technology from Analog Research grant NNX16AL08G (RC) Author contributions: Conceptualization: AFD, AP, GB, MRW, RC Methodology: AFD, AP, GB, MRW, RC Investigation: AFD, AP, GB, MRW, RC Visualization: AFD, AP, GB, RC Funding acquisition: MRW, RC Project administration: MRW, RC Supervision: MRW, RC Writing - original draft: AFD, AP, GB, MRW, RC Writing - review & editing: AFD, AP, GB, MRW, RC Competing interests: Authors declare that they have no competing interests. Data and materials availability: All data are available in the main text or the supplementary materials. NOTE: This is the author's version of the work. It is posted here by permission of the AAAS for personal use, not for redistribution. The definitive version was published in Science Robotics on 23 Aug 2023, DOI: 10.1126/scirobotics.adi5227.

This thesis focuses on improving the pre-training of natural language models using unsupervised raw data to make them more efficient and aligned with downstream applications. In the first part, we introduce three alternative pre-training objectives to BERT's Masked Language Modeling (MLM), namely Random Token Substitution (RTS), Cluster-based Random Token Substitution (C-RTS), and Swapped Language Modeling (SLM). These objectives involve token swapping instead of masking, with RTS and C-RTS aiming to predict token originality and SLM predicting the original token values. Results show that RTS and C-RTS require less pre-training time while maintaining performance comparable to MLM. Surprisingly, SLM outperforms MLM on certain tasks despite using the same computational budget. In the second part, we proposes self-supervised pre-training tasks that align structurally with downstream applications, reducing the need for labeled data. We use large corpora like Wikipedia and CC-News to train models to recognize if text spans originate from the same paragraph or document in several ways. By doing continuous pre-training, starting from existing models like RoBERTa, ELECTRA, DeBERTa, BART, and T5, we demonstrate significant performance improvements in tasks like Fact Verification, Answer Sentence Selection, and Summarization. These improvements are especially pronounced when limited annotation data is available. The proposed objectives also achieve state-of-the-art results on various benchmark datasets, including FEVER (dev set), ASNQ, WikiQA, and TREC-QA, as well as enhancing the quality of summaries. Importantly, these techniques can be easily integrated with other methods without altering the internal structure of Transformer models, making them versatile for various NLP applications.

When the Titan submersible, carrying five sightseers to the wreck of the Titanic, blew up thousands of metres under the ocean surface in June, it underscored why humanity knows more about the surface of some other planets than about the depths of the Earth's oceans. Oceans cover more than 70 percent of the earth's surface. Yet, this underwater world is a challenging place to explore, as the Titan disaster showed. The deepest point under water, the Challenger Deep in the Pacific Ocean, is 11,000 metres deep, more than the height of Mount Everest. The light doesn't penetrate to such depths.

In this paper, we present algorithms to identify environmental hotspots using mobile sensors. We examine two approaches: one involving a single robot and another using multiple robots coordinated through a decentralized robot system. We introduce an adaptive algorithm that does not require precise knowledge of Gaussian Processes (GPs) hyperparameters, making the modeling process more flexible. The robots operate for a pre-defined time in the environment. The multi-robot system uses Voronoi partitioning to divide tasks and a Monte Carlo Tree Search for optimal path planning. Our tests on synthetic and a real-world dataset of Chlorophyll density from a Pacific Ocean sub-region suggest that accurate estimation of GP hyperparameters may not be essential for hotspot detection, potentially simplifying environmental monitoring tasks.

Large language models show improved downstream task performance when prompted to generate step-by-step reasoning to justify their final answers. These reasoning steps greatly improve model interpretability and verification, but objectively studying their correctness (independent of the final answer) is difficult without reliable methods for automatic evaluation. We simply do not know how often the stated reasoning steps actually support the final end task predictions. In this work, we present ROSCOE, a suite of interpretable, unsupervised automatic scores that improve and extend previous text generation evaluation metrics. To evaluate ROSCOE against baseline metrics, we design a typology of reasoning errors and collect synthetic and human evaluation scores on commonly used reasoning datasets. In contrast with existing metrics, ROSCOE can measure semantic consistency, logicality, informativeness, fluency, and factuality - among other traits - by leveraging properties of step-by-step rationales. We empirically verify the strength of our metrics on five human annotated and six programmatically perturbed diagnostics datasets - covering a diverse set of tasks that require reasoning skills and show that ROSCOE can consistently outperform baseline metrics.

This research focuses on the estimation of a non-parametric regression function designed for data with simultaneous time and space dependencies. In such a context, we study the Trend Filtering, a nonparametric estimator introduced by \cite{mammen1997locally} and \cite{rudin1992nonlinear}. For univariate settings, the signals we consider are assumed to have a kth weak derivative with bounded total variation, allowing for a general degree of smoothness. In the multivariate scenario, we study a $K$-Nearest Neighbor fused lasso estimator as in \cite{padilla2018adaptive}, employing an ADMM algorithm, suitable for signals with bounded variation that adhere to a piecewise Lipschitz continuity criterion. By aligning with lower bounds, the minimax optimality of our estimators is validated. A unique phase transition phenomenon, previously uncharted in Trend Filtering studies, emerges through our analysis. Both Simulation studies and real data applications underscore the superior performance of our method when compared with established techniques in the existing literature.

Executives of top tech firms, including Google and Intel, have met with business leaders in Vietnam as part of United States President Joe Biden's landmark visit to the Southeast Asian country. Tech leaders joined Biden and US Secretary of State Antony Blinken on Monday for an "innovation and investment summit" attended by Vietnamese firms, including electric car maker VinFast, internet company VNG and digital wallet provider Momo. Washington and Hanoi are seeking to deepen their cooperation amid shared concerns about China's rising power and influence. The US views Vietnam as a key plank of its plans to reduce its reliance on China for strategic resources, such as semiconductors and rare earth minerals. Vietnam has territorial disputes with Beijing in the South China Sea.

Machine Reading Comprehension (MRC) models tend to take advantage of spurious correlations (also known as dataset bias or annotation artifacts in the research community). Consequently, these models may perform the MRC task without fully comprehending the given context and question, which is undesirable since it may result in low robustness against distribution shift. The main focus of this paper is answer-position bias, where a significant percentage of training questions have answers located solely in the first sentence of the context. We propose a Single-Sentence Reader as a new approach for addressing answer position bias in MRC. Remarkably, in our experiments with six different models, our proposed Single-Sentence Readers trained on biased dataset achieve results that nearly match those of models trained on normal dataset, proving their effectiveness in addressing the answer position bias. Our study also discusses several challenges our Single-Sentence Readers encounter and proposes a potential solution.

Predicting vehicle trajectories is crucial for ensuring automated vehicle operation efficiency and safety, particularly on congested multi-lane highways. In such dynamic environments, a vehicle's motion is determined by its historical behaviors as well as interactions with surrounding vehicles. These intricate interactions arise from unpredictable motion patterns, leading to a wide range of driving behaviors that warrant in-depth investigation. This study presents the Graph-based Interaction-aware Multi-modal Trajectory Prediction (GIMTP) framework, designed to probabilistically predict future vehicle trajectories by effectively capturing these interactions. Within this framework, vehicles' motions are conceptualized as nodes in a time-varying graph, and the traffic interactions are represented by a dynamic adjacency matrix. To holistically capture both spatial and temporal dependencies embedded in this dynamic adjacency matrix, the methodology incorporates the Diffusion Graph Convolutional Network (DGCN), thereby providing a graph embedding of both historical states and future states. Furthermore, we employ a driving intention-specific feature fusion, enabling the adaptive integration of historical and future embeddings for enhanced intention recognition and trajectory prediction. This model gives two-dimensional predictions for each mode of longitudinal and lateral driving behaviors and offers probabilistic future paths with corresponding probabilities, addressing the challenges of complex vehicle interactions and multi-modality of driving behaviors. Validation using real-world trajectory datasets demonstrates the efficiency and potential.