Large-scale generative language and vision-language models (LLMs and VLMs) excel in few-shot in-context learning for decision making and instruction following. However, they require high-quality exemplar demonstrations to be included in their context window. In this work, we ask: Can LLMs and VLMs generate their own prompt examples from generic, sub-optimal demonstrations? We propose In-Context Abstraction Learning (ICAL), a method that builds a memory of multimodal experience insights from sub-optimal demonstrations and human feedback. Given a noisy demonstration in a new domain, VLMs abstract the trajectory into a general program by fixing inefficient actions and annotating cognitive abstractions: task relationships, object state changes, temporal subgoals, and task construals. These abstractions are refined and adapted interactively through human feedback while the agent attempts to execute the trajectory in a similar environment. The resulting abstractions, when used as exemplars in the prompt, significantly improve decision-making in retrieval-augmented LLM and VLM agents. Our ICAL agent surpasses the state-of-the-art in dialogue-based instruction following in TEACh, multimodal web agents in VisualWebArena, and action anticipation in Ego4D. In TEACh, we achieve a 12.6% improvement in goal-condition success. In VisualWebArena, our task success rate improves over the SOTA from 14.3% to 22.7%. In Ego4D action forecasting, we improve over few-shot GPT-4V and remain competitive with supervised models. We show finetuning our retrieval-augmented in-context agent yields additional improvements. Our approach significantly reduces reliance on expert-crafted examples and consistently outperforms in-context learning from action plans that lack such insights.

In the rapidly evolving landscape of the IoT, the security of connected devices has become a paramount concern. This paper explores the concept of proactive threat hunting as a pivotal strategy for enhancing the security and sustainability of IoT systems. Proactive threat hunting is an alternative to traditional reactive security measures that analyses IoT networks continuously and in advance to find and eliminate threats before they occure. By improving the security posture of IoT devices this approach significantly contributes to extending IoT operational lifespan and reduces environmental impact. By integrating security metrics similar to the Common Vulnerability Scoring System (CVSS) into consumer platforms, this paper argues that proactive threat hunting can elevate user awareness about the security of IoT devices. This has the potential to impact consumer choices and encourage a security-conscious mindset in both the manufacturing and user communities. Through a comprehensive analysis, this study demonstrates how proactive threat hunting can contribute to the development of a more secure, sustainable, and user-aware IoT ecosystem.

Evaluating aligned large language models' (LLMs) ability to recognize and reject unsafe user requests is crucial for safe, policy-compliant deployments. Existing evaluation efforts, however, face three limitations that we address with SORRY-Bench, our proposed benchmark. First, existing methods often use coarse-grained taxonomies of unsafe topics, and are over-representing some fine-grained topics. For example, among the ten existing datasets that we evaluated, tests for refusals of self-harm instructions are over 3x less represented than tests for fraudulent activities. SORRY-Bench improves on this by using a fine-grained taxonomy of 45 potentially unsafe topics, and 450 class-balanced unsafe instructions, compiled through human-in-the-loop methods. Second, linguistic characteristics and formatting of prompts are often overlooked, like different languages, dialects, and more -- which are only implicitly considered in many evaluations. We supplement SORRY-Bench with 20 diverse linguistic augmentations to systematically examine these effects. Third, existing evaluations rely on large LLMs (e.g., GPT-4) for evaluation, which can be computationally expensive. We investigate design choices for creating a fast, accurate automated safety evaluator. By collecting 7K+ human annotations and conducting a meta-evaluation of diverse LLM-as-a-judge designs, we show that fine-tuned 7B LLMs can achieve accuracy comparable to GPT-4 scale LLMs, with lower computational cost. Putting these together, we evaluate over 40 proprietary and open-source LLMs on SORRY-Bench, analyzing their distinctive refusal behaviors. We hope our effort provides a building block for systematic evaluations of LLMs' safety refusal capabilities, in a balanced, granular, and efficient manner.

Global Station Weather Forecasting (GSWF) is crucial for various sectors, including aviation, agriculture, energy, and disaster preparedness. Recent advancements in deep learning have significantly improved the accuracy of weather predictions by optimizing models based on public meteorological data. However, existing public datasets for GSWF optimization and benchmarking still suffer from significant limitations, such as small sizes, limited temporal coverage, and a lack of comprehensive variables. These shortcomings prevent them from effectively reflecting the benchmarks of current forecasting methods and fail to support the real needs of operational weather forecasting. To address these challenges, we present the WEATHER-5K dataset. This dataset comprises a comprehensive collection of data from 5,672 weather stations worldwide, spanning a 10-year period with one-hour intervals. It includes multiple crucial weather elements, providing a more reliable and interpretable resource for forecasting. Furthermore, our WEATHER-5K dataset can serve as a benchmark for comprehensively evaluating existing well-known forecasting models, extending beyond GSWF methods to support future time-series research challenges and opportunities.

Retrieval-augmented generation (RAG) has received much attention for Open-domain question-answering (ODQA) tasks as a means to compensate for the parametric knowledge of large language models (LLMs). While previous approaches focused on processing retrieved passages to remove irrelevant context, they still rely heavily on the quality of retrieved passages which can degrade if the question is ambiguous or complex. In this paper, we propose a simple yet efficient method called question and passage augmentation via LLMs for open-domain QA. Our method first decomposes the original questions into multiple-step sub-questions. By augmenting the original question with detailed sub-questions and planning, we are able to make the query more specific on what needs to be retrieved, improving the retrieval performance. In addition, to compensate for the case where the retrieved passages contain distracting information or divided opinions, we augment the retrieved passages with self-generated passages by LLMs to guide the answer extraction. Experimental results show that the proposed scheme outperforms the previous state-of-the-art and achieves significant performance gain over existing RAG methods.

We use (multi)modal deep neural networks (DNNs) to probe for sites of multimodal integration in the human brain by predicting stereoencephalography (SEEG) recordings taken while human subjects watched movies. We operationalize sites of multimodal integration as regions where a multimodal vision-language model predicts recordings better than unimodal language, unimodal vision, or linearly-integrated language-vision models. Our target DNN models span different architectures (e.g., convolutional networks and transformers) and multimodal training techniques (e.g., cross-attention and contrastive learning). As a key enabling step, we first demonstrate that trained vision and language models systematically outperform their randomly initialized counterparts in their ability to predict SEEG signals. We then compare unimodal and multimodal models against one another. Because our target DNN models often have different architectures, number of parameters, and training sets (possibly obscuring those differences attributable to integration), we carry out a controlled comparison of two models (SLIP and SimCLR), which keep all of these attributes the same aside from input modality. Using this approach, we identify a sizable number of neural sites (on average 141 out of 1090 total sites or 12.94%) and brain regions where multimodal integration seems to occur. Additionally, we find that among the variants of multimodal training techniques we assess, CLIP-style training is the best suited for downstream prediction of the neural activity in these sites.

The ability to manipulate logical-mathematical symbols (LMS), encompassing tasks such as calculation, reasoning, and programming, is a cognitive skill arguably unique to humans. Considering the relatively recent emergence of this ability in human evolutionary history, it has been suggested that LMS processing may build upon more fundamental cognitive systems, possibly through neuronal recycling. Previous studies have pinpointed two primary candidates, natural language processing and spatial cognition. Existing comparisons between these domains largely relied on task-level comparison, which may be confounded by task idiosyncrasy. The present study instead compared the neural correlates at the domain level with both automated meta-analysis and synthesized maps based on three representative LMS tasks, reasoning, calculation, and mental programming. Our results revealed a more substantial cortical overlap between LMS processing and spatial cognition, in contrast to language processing. Furthermore, in regions activated by both spatial and language processing, the multivariate activation pattern for LMS processing exhibited greater multivariate similarity to spatial cognition than to language processing. A hierarchical clustering analysis further indicated that typical LMS tasks were indistinguishable from spatial cognition tasks at the neural level, suggesting an inherent connection between these two cognitive processes. Taken together, our findings support the hypothesis that spatial cognition is likely the basis of LMS processing, which may shed light on the limitations of large language models in logical reasoning, particularly those trained exclusively on textual data without explicit emphasis on spatial content.

Artificial Intelligence (AI) has the potential to revolutionize space exploration by delegating several spacecraft decisions to an onboard AI instead of relying on ground control and predefined procedures. It is likely that there will be an AI/ML Processing Unit onboard the spacecraft running an inference engine. The neural-network will have pre-installed parameters that can be updated onboard by uploading, by telecommands, parameters obtained by training on the ground. However, satellite uplinks have limited bandwidth and transmissions can be costly. Furthermore, a mission operating with a suboptimal neural network will miss out on valuable scientific data. Smaller networks can thereby decrease the uplink cost, while increasing the value of the scientific data that is downloaded. In this work, we evaluate and discuss the use of reduced-precision and bare-minimum neural networks to reduce the time for upload. As an example of an AI use case, we focus on the NASA's Magnetosperic MultiScale (MMS) mission. We show how an AI onboard could be used in the Earth's magnetosphere to classify data to selectively downlink higher value data or to recognize a region-of-interest to trigger a burst-mode, collecting data at a high-rate. Using a simple filtering scheme and algorithm, we show how the start and end of a region-of-interest can be detected in on a stream of classifications. To provide the classifications, we use an established Convolutional Neural Network (CNN) trained to an accuracy >94%. We also show how the network can be reduced to a single linear layer and trained to the same accuracy as the established CNN. Thereby, reducing the overall size of the model by up to 98.9%. We further show how each network can be reduced by up to 75% of its original size, by using lower-precision formats to represent the network parameters, with a change in accuracy of less than 0.6 percentage points.

Federated learning is highly susceptible to model poisoning attacks, especially those meticulously crafted for servers. Traditional defense methods mainly focus on updating assessments or robust aggregation against manually crafted myopic attacks. When facing advanced attacks, their defense stability is notably insufficient. Therefore, it is imperative to develop adaptive defenses against such advanced poisoning attacks. We find that benign clients exhibit significantly higher data distribution stability than malicious clients in federated learning in both CV and NLP tasks. Therefore, the malicious clients can be recognized by observing the stability of their data distribution. In this paper, we propose AdaAggRL, an RL-based Adaptive Aggregation method, to defend against sophisticated poisoning attacks. Specifically, we first utilize distribution learning to simulate the clients' data distributions. Then, we use the maximum mean discrepancy (MMD) to calculate the pairwise similarity of the current local model data distribution, its historical data distribution, and global model data distribution. Finally, we use policy learning to adaptively determine the aggregation weights based on the above similarities. Experiments on four real-world datasets demonstrate that the proposed defense model significantly outperforms widely adopted defense models for sophisticated attacks.

Large language models can now generate political messages as persuasive as those written by humans, raising concerns about how far this persuasiveness may continue to increase with model size. Here, we generate 720 persuasive messages on 10 U.S. political issues from 24 language models spanning several orders of magnitude in size. We then deploy these messages in a large-scale randomized survey experiment (N = 25,982) to estimate the persuasive capability of each model. Our findings are twofold. First, we find evidence of a log scaling law: model persuasiveness is characterized by sharply diminishing returns, such that current frontier models are barely more persuasive than models smaller in size by an order of magnitude or more. Second, mere task completion (coherence, staying on topic) appears to account for larger models' persuasive advantage. These findings suggest that further scaling model size will not much increase the persuasiveness of static LLM-generated messages.