Selecting high-quality pre-training data is important for creating capable language models, but existing methods rely on simple heuristics. We introduce QuRating, a method for selecting pre-training data that captures the abstract qualities of texts which humans intuitively perceive. In this paper, we investigate four qualities - writing style, required expertise, facts & trivia, and educational value. We find that LLMs are able to discern these qualities and observe that they are better at making pairwise judgments of texts than at rating the quality of a text directly. We train a QuRater model to learn scalar ratings from pairwise judgments, and use it to annotate a 260B training corpus with quality ratings for each of the four criteria. In our experiments, we select 30B tokens according to the different quality ratings and train 1.3B-parameter language models on the selected data. We find that it is important to balance quality and diversity, as selecting only the highest-rated documents leads to poor results. When we sample using quality ratings as logits over documents, our models achieve lower perplexity and stronger in-context learning performance than baselines. Beyond data selection, we use the quality ratings to construct a training curriculum which improves performance without changing the training dataset. We extensively analyze the quality ratings and discuss their characteristics, biases, and wider implications.

With the development of audio playback devices and fast data transmission, the demand for high sound quality is rising, for both entertainment and communications. In this quest for better sound quality, challenges emerge from distortions and interferences originating at the recording side or caused by an imperfect transmission pipeline. To address this problem, audio restoration methods aim to recover clean sound signals from the corrupted input data. We present here audio restoration algorithms based on diffusion models, with a focus on speech enhancement and music restoration tasks. Traditional approaches, often grounded in handcrafted rules and statistical heuristics, have shaped our understanding of audio signals. In the past decades, there has been a notable shift towards data-driven methods that exploit the modeling capabilities of deep neural networks (DNNs). Deep generative models, and among them diffusion models, have emerged as powerful techniques for learning complex data distributions. However, relying solely on DNN-based learning approaches carries the risk of reducing interpretability, particularly when employing end-to-end models. Nonetheless, data-driven approaches allow more flexibility in comparison to statistical model-based frameworks whose performance depends on distributional and statistical assumptions that can be difficult to guarantee. Here, we aim to show that diffusion models can combine the best of both worlds and offer the opportunity to design audio restoration algorithms with a good degree of interpretability and a remarkable performance in terms of sound quality.

In many collaborative settings, artificial intelligence (AI) agents must be able to adapt to new teammates that use unknown or previously unobserved strategies. While often simple for humans, this can be challenging for AI agents. For example, if an AI agent learns to drive alongside others (a training set) that only drive on one side of the road, it may struggle to adapt this experience to coordinate with drivers on the opposite side, even if their behaviours are simply flipped along the left-right symmetry. To address this we introduce symmetry-breaking augmentations (SBA), which increases diversity in the behaviour of training teammates by applying a symmetry-flipping operation. By learning a best-response to the augmented set of teammates, our agent is exposed to a wider range of behavioural conventions, improving performance when deployed with novel teammates. We demonstrate this experimentally in two settings, and show that our approach improves upon previous ad hoc teamwork results in the challenging card game Hanabi. We also propose a general metric for estimating symmetry-dependency amongst a given set of policies.

The rapid advancement of Large Language Models (LLMs) has demonstrated their vast potential across various domains, attributed to their extensive pretraining knowledge and exceptional generalizability. However, LLMs often encounter challenges in generating harmful content when faced with problematic prompts. To address this problem, existing work attempted to implement a gradient ascent based approach to prevent LLMs from producing harmful output. While these methods can be effective, they frequently impact the model utility in responding to normal prompts. To address this gap, we introduce Selective Knowledge negation Unlearning (SKU), a novel unlearning framework for LLMs, designed to eliminate harmful knowledge while preserving utility on normal prompts. Specifically, SKU is consisted of two stages: harmful knowledge acquisition stage and knowledge negation stage. The first stage aims to identify and acquire harmful knowledge within the model, whereas the second is dedicated to remove this knowledge. SKU selectively isolates and removes harmful knowledge in model parameters, ensuring the model's performance remains robust on normal prompts. Our experiments conducted across various LLM architectures demonstrate that SKU identifies a good balance point between removing harmful information and preserving utility.

Global ambient air pollution, a transboundary challenge, is typically addressed through interventions relying on data from spatially sparse and heterogeneously placed monitoring stations. These stations often encounter temporal data gaps due to issues such as power outages. In response, we have developed a scalable, data-driven, supervised machine learning framework. This model is designed to impute missing temporal and spatial measurements, thereby generating a comprehensive dataset for pollutants including NO$_2$, O$_3$, PM$_{10}$, PM$_{2.5}$, and SO$_2$. The dataset, with a fine granularity of 0.25$^{\circ}$ at hourly intervals and accompanied by prediction intervals for each estimate, caters to a wide range of stakeholders relying on outdoor air pollution data for downstream assessments. This enables more detailed studies. Additionally, the model's performance across various geographical locations is examined, providing insights and recommendations for strategic placement of future monitoring stations to further enhance the model's accuracy.

Neglecting complex aerodynamic effects hinders high-speed yet high-precision multirotor autonomy. In this paper, we present a computationally efficient learning-based model predictive controller that simultaneously optimizes a trajectory that can be tracked within the physical limits (on thrust and orientation) of the multirotor system despite unknown aerodynamic forces and adapts the control input. To do this, we leverage the well-known differential flatness property of multirotors, which allows us to transform their nonlinear dynamics into a linear model. The main limitation of current flatness-based planning and control approaches is that they often neglect dynamic feasibility. This is because these constraints are nonlinear as a result of the mapping between the input, i.e., multirotor thrust, and the flat state. In our approach, we learn a novel representation of the drag forces by learning the mapping from the flat state to the multirotor thrust vector (in a world frame) as a Gaussian Process (GP). Our proposed approach leverages the properties of GPs to develop a convex optimal controller that can be iteratively solved as a second-order cone program (SOCP). In simulation experiments, our proposed approach outperforms related model predictive controllers that do not account for aerodynamic effects on trajectory feasibility, leading to a reduction of up to 55% in absolute tracking error.

Robotic manipulation of deformable linear objects (DLOs) is an active area of research, though emerging applications, like automotive wire harness installation, introduce constraints that have not been considered in prior work. Confined workspaces and limited visibility complicate prior assumptions of multi-robot manipulation and direct measurement of DLO configuration (state). This work focuses on single-arm manipulation of stiff DLOs (StDLOs) connected to form a DLO network (DLON), for which the measurements (output) are the endpoint poses of the DLON, which are subject to unknown dynamics during manipulation. To demonstrate feasibility of output-based control without state estimation, direct input-output dynamics are shown to exist by training neural network models on simulated trajectories. Output dynamics are then approximated with polynomials and found to contain well-known rigid body dynamics terms. A composite model consisting of a rigid body model and an online data-driven residual is developed, which predicts output dynamics more accurately than either model alone, and without prior experience with the system. An adaptive model predictive controller is developed with the composite model for DLON manipulation, which completes DLON installation tasks, both in simulation and with a physical automotive wire harness.

Heterogeneous autonomous robot teams consisting of multirotor and uncrewed surface vessels (USVs) have the potential to enable various maritime applications, including advanced search-and-rescue operations. A critical requirement of these applications is the ability to land a multirotor on a USV for tasks such as recharging. This paper addresses the challenge of safely landing a multirotor on a cooperative USV in harsh open waters. To tackle this problem, we propose a novel sequential distributed model predictive control (MPC) scheme for cooperative multirotor-USV landing. Our approach combines standard tracking MPCs for the multirotor and USV with additional artificial intermediate goal locations. These artificial goals enable the robots to coordinate their cooperation without prior guidance. Each vehicle solves an individual optimization problem for both the artificial goal and an input that tracks it but only communicates the former to the other vehicle. The artificial goals are penalized by a suitable coupling cost. Furthermore, our proposed distributed MPC scheme utilizes a spatial-temporal wave model to coordinate in real-time a safer landing location and time the multirotor's landing to limit severe tilt of the USV.

The report describes the discussions from the Workshop on Mathematical Opportunities in Digital Twins (MATH-DT) from December 11-13, 2023, George Mason University. It illustrates that foundational Mathematical advances are required for Digital Twins (DTs) that are different from traditional approaches. A traditional model, in biology, physics, engineering or medicine, starts with a generic physical law (e.g., equations) and is often a simplification of reality. A DT starts with a specific ecosystem, object or person (e.g., personalized care) representing reality, requiring multi -scale, -physics modeling and coupling. Thus, these processes begin at opposite ends of the simulation and modeling pipeline, requiring different reliability criteria and uncertainty assessments. Additionally, unlike existing approaches, a DT assists humans to make decisions for the physical system, which (via sensors) in turn feeds data into the DT, and operates for the life of the physical system. While some of the foundational mathematical research can be done without a specific application context, one must also keep specific applications in mind for DTs. E.g., modeling a bridge or a biological system (a patient), or a socio-technical system (a city) is very different. The models range from differential equations (deterministic/uncertain) in engineering, to stochastic in biology, including agent-based. These are multi-scale hybrid models or large scale (multi-objective) optimization problems under uncertainty. There are no universal models or approaches. For e.g., Kalman filters for forecasting might work in engineering, but can fail in biomedical domain. Ad hoc studies, with limited systematic work, have shown that AI/ML methods can fail for simple engineering systems and can work well for biomedical problems. A list of `Mathematical Opportunities and Challenges' concludes the report.

Transmission system operators have a growing need for more accurate forecasting of electricity demand. Current electricity systems largely require demand forecasting so that the electricity market establishes electricity prices as well as the programming of production units. The companies that are part of the electrical system use exclusive software to obtain predictions, based on the use of time series and prediction tools, whether statistical or artificial intelligence. However, the most common form of prediction is based on hybrid models that use both technologies. In any case, it is software with a complicated structure, with a large number of associated variables and that requires a high computational load to make predictions. The predictions they can offer are not much better than those that simple models can offer. In this paper we present a MATLAB toolbox created for the prediction of electrical demand. The toolbox implements multiple seasonal Holt-Winters exponential smoothing models and neural network models. The models used include the use of discrete interval mobile seasonalities (DIMS) to improve forecasting on special days. Additionally, the results of its application in various electrical systems in Europe are shown, where the results obtained can be seen. The use of this library opens a new avenue of research for the use of models with discrete and complex seasonalities in other fields of application.