Large language models (LLMs) trained on general domain corpora showed remarkable results on natural language processing (NLP) tasks. However, previous research demonstrated LLMs trained using domain-focused corpora perform better on specialized tasks. Inspired by this pivotal insight, we developed INDUS, a comprehensive suite of LLMs tailored for the Earth science, biology, physics, heliophysics, planetary sciences and astrophysics domains and trained using curated scientific corpora drawn from diverse data sources. The suite of models include: (1) an encoder model trained using domain-specific vocabulary and corpora to address natural language understanding tasks, (2) a contrastive-learning-based general text embedding model trained using a diverse set of datasets drawn from multiple sources to address information retrieval tasks and (3) smaller versions of these models created using knowledge distillation techniques to address applications which have latency or resource constraints. We also created three new scientific benchmark datasets namely, CLIMATE-CHANGE-NER (entity-recognition), NASA-QA (extractive QA) and NASA-IR (IR) to accelerate research in these multi-disciplinary fields. Finally, we show that our models outperform both general-purpose encoders (RoBERTa) and existing domain-specific encoders (SciBERT) on these new tasks as well as existing benchmark tasks in the domains of interest.

As public sector agencies rapidly introduce new AI tools in high-stakes domains like social services, it becomes critical to understand how decisions to adopt these tools are made in practice. We borrow from the anthropological practice to ``study up'' those in positions of power, and reorient our study of public sector AI around those who have the power and responsibility to make decisions about the role that AI tools will play in their agency. Through semi-structured interviews and design activities with 16 agency decision-makers, we examine how decisions about AI design and adoption are influenced by their interactions with and assumptions about other actors within these agencies (e.g., frontline workers and agency leaders), as well as those above (legal systems and contracted companies), and below (impacted communities). By centering these networks of power relations, our findings shed light on how infrastructural, legal, and social factors create barriers and disincentives to the involvement of a broader range of stakeholders in decisions about AI design and adoption. Agency decision-makers desired more practical support for stakeholder involvement around public sector AI to help overcome the knowledge and power differentials they perceived between them and other stakeholders (e.g., frontline workers and impacted community members). Building on these findings, we discuss implications for future research and policy around actualizing participatory AI approaches in public sector contexts.

Federated learning (FL) is a decentralized learning technique that enables participating devices to collaboratively build a shared Machine Leaning (ML) or Deep Learning (DL) model without revealing their raw data to a third party. Due to its privacy-preserving nature, FL has sparked widespread attention for building Intrusion Detection Systems (IDS) within the realm of cybersecurity. However, the data heterogeneity across participating domains and entities presents significant challenges for the reliable implementation of an FL-based IDS. In this paper, we propose an effective method called Statistical Averaging (StatAvg) to alleviate non-independently and identically (non-iid) distributed features across local clients' data in FL. In particular, StatAvg allows the FL clients to share their individual data statistics with the server, which then aggregates this information to produce global statistics. The latter are shared with the clients and used for universal data normalisation. It is worth mentioning that StatAvg can seamlessly integrate with any FL aggregation strategy, as it occurs before the actual FL training process. The proposed method is evaluated against baseline approaches using datasets for network and host Artificial Intelligence (AI)-powered IDS. The experimental results demonstrate the efficiency of StatAvg in mitigating non-iid feature distributions across the FL clients compared to the baseline methods.

This paper studies the challenge of developing robots capable of understanding under-specified instructions for creating functional object arrangements, such as "set up a dining table for two"; previous arrangement approaches have focused on much more explicit instructions, such as "put object A on the table." We introduce a framework, SetItUp, for learning to interpret under-specified instructions. SetItUp takes a small number of training examples and a human-crafted program sketch to uncover arrangement rules for specific scene types. By leveraging an intermediate graph-like representation of abstract spatial relationships among objects, SetItUp decomposes the arrangement problem into two subproblems: i) learning the arrangement patterns from limited data and ii) grounding these abstract relationships into object poses. SetItUp leverages large language models (LLMs) to propose the abstract spatial relationships among objects in novel scenes as the constraints to be satisfied; then, it composes a library of diffusion models associated with these abstract relationships to find object poses that satisfy the constraints. We validate our framework on a dataset comprising study desks, dining tables, and coffee tables, with the results showing superior performance in generating physically plausible, functional, and aesthetically pleasing object arrangements compared to existing models.

Sensor placement optimization methods have been studied extensively. They can be applied to a wide range of applications, including surveillance of known environments, optimal locations for 5G towers, and placement of missile defense systems. However, few works explore the robustness and efficiency of the resulting sensor network concerning sensor failure or adversarial attacks. This paper addresses this issue by optimizing for the least number of sensors to achieve multiple coverage of non-simply connected domains by a prescribed number of sensors. We introduce a new objective function for the greedy (next-best-view) algorithm to design efficient and robust sensor networks and derive theoretical bounds on the network's optimality. We further introduce a Deep Learning model to accelerate the algorithm for near real-time computations. The Deep Learning model requires the generation of training examples. Correspondingly, we show that understanding the geometric properties of the training data set provides important insights into the performance and training process of deep learning techniques. Finally, we demonstrate that a simple parallel version of the greedy approach using a simpler objective can be highly competitive.

Controlling contact forces during interactions is critical for locomotion and manipulation tasks. While sim-to-real reinforcement learning (RL) has succeeded in many contact-rich problems, current RL methods achieve forceful interactions implicitly without explicitly regulating forces. We propose a method for training RL policies for direct force control without requiring access to force sensing. We showcase our method on a whole-body control platform of a quadruped robot with an arm. Such force control enables us to perform gravity compensation and impedance control, unlocking compliant whole-body manipulation. The learned whole-body controller with variable compliance makes it intuitive for humans to teleoperate the robot by only commanding the manipulator, and the robot's body adjusts automatically to achieve the desired position and force. Consequently, a human teleoperator can easily demonstrate a wide variety of loco-manipulation tasks. To the best of our knowledge, we provide the first deployment of learned whole-body force control in legged manipulators, paving the way for more versatile and adaptable legged robots.

Collective intelligence of diverse groups is key for tackling many of today's grand challenges such as fostering resilience and climate adaptation. Information exchange across such diverse groups is crucial for collective intelligence, especially in volatile environments. To facilitate inter-group information exchange, Informational Boundary Spanners (IBSs) as pivotal information exchange 'hubs' are promising. However, the mechanisms that drive the emergence of IBSs remain poorly understood. To address this gap there is first a need for a method to identify and measure the emergence of IBSs. Second, an Agent-Based Modelling (ABM) framework is not available to systematically study mechanisms for the emergence of IBSs in volatile environments. Third, even though the ability to learn who provides high-quality information is thought to be essential to explain the emergence of IBSs, a rigorous test of this mechanism is missing. The learning mechanism is formalized using an ABM framework, with the model's outputs analyzed using the proposed IBS emergence measurement method. To illustrate both the method and the learning mechanism, we present a case study focused on information sharing in the volatile environment of a disaster. The study shows that learning constitutes a mechanism for the emergence of effective IBSs in (a) low-volatility environments characterised by low uncertainty and (b) in high-volatility environments characterised by rapid change if the number of inter-group connections is sufficient. With the method and model, this paper aims to lay the foundations for exploring mechanisms for the emergence of IBSs that facilitate inter-group information exchange. This article advances collective intelligence by providing the essential elements for measuring and understanding the emergence of IBSs and exploring the effect of learning on their emergence in volatile environments.

Embedding news articles is a crucial tool for multiple fields, such as media bias detection, identifying fake news, and news recommendations. However, existing news embedding methods are not optimized for capturing the latent context of news events. In many cases, news embedding methods rely on full-textual information and neglect the importance of time-relevant embedding generation. Here, we aim to address these shortcomings by presenting a novel lightweight method that optimizes news embedding generation by focusing on the entities and themes mentioned in the articles and their historical connections to specific events. We suggest a method composed of three stages. First, we process and extract the events, entities, and themes for the given news articles. Second, we generate periodic time embeddings for themes and entities by training timely separated GloVe models on current and historical data. Lastly, we concatenate the news embeddings generated by two distinct approaches: Smooth Inverse Frequency (SIF) for article-level vectors and Siamese Neural Networks for embeddings with nuanced event-related information. To test and evaluate our method, we leveraged over 850,000 news articles and 1,000,000 events from the GDELT project. For validation purposes, we conducted a comparative analysis of different news embedding generation methods, applying them twice to a shared event detection task - first on articles published within the same day and subsequently on those published within the same month. Our experiments show that our method significantly improves the Precision-Recall (PR) AUC across all tasks and datasets. Specifically, we observed an average PR AUC improvement of 2.15% and 2.57% compared to SIF, as well as 2.57% and 2.43% compared to the semi-supervised approach for daily and monthly shared event detection tasks, respectively.

For many space applications, traditional control methods are often used during operation. However, as the number of space assets continues to grow, autonomous operation can enable rapid development of control methods for different space related tasks. One method of developing autonomous control is Reinforcement Learning (RL), which has become increasingly popular after demonstrating promising performance and success across many complex tasks. While it is common for RL agents to learn bounded continuous control values, this may not be realistic or practical for many space tasks that traditionally prefer an on/off approach for control. This paper analyzes using discrete action spaces, where the agent must choose from a predefined list of actions. The experiments explore how the number of choices provided to the agents affects their measured performance during and after training. This analysis is conducted for an inspection task, where the agent must circumnavigate an object to inspect points on its surface, and a docking task, where the agent must move into proximity of another spacecraft and "dock" with a low relative speed. A common objective of both tasks, and most space tasks in general, is to minimize fuel usage, which motivates the agent to regularly choose an action that uses no fuel. Our results show that a limited number of discrete choices leads to optimal performance for the inspection task, while continuous control leads to optimal performance for the docking task.

Guardrails to prevent artificial intelligence models behind chatbots from issuing illegal, toxic or explicit responses can be bypassed with simple techniques, UK government researchers have found. The UK's AI Safety Institute (AISI) said systems it had tested were "highly vulnerable" to jailbreaks, a term for text prompts designed to elicit a response that a model is supposedly trained to avoid issuing. The AISI said it had tested five unnamed large language models (LLM) – the technology that underpins chatbots – and circumvented their safeguards with relative ease, even without concerted attempts to beat their guardrails. "All tested LLMs remain highly vulnerable to basic jailbreaks, and some will provide harmful outputs even without dedicated attempts to circumvent their safeguards," wrote AISI researchers in an update on their testing regime. The AISI found that safeguards could be circumvented with "relatively simple" attacks, by, for instance, instructing the system to start its response with phrases like "Sure, I'm happy to help".