At its core, machine learning seeks to train models that reliably generalize beyond noisy observations; however, the theoretical vacuum in which state-of-the-art universal approximation theorems (UATs) operate isolates them from this goal, as they assume noiseless data and allow network parameters to be chosen freely, independent of algorithmic realism. This paper bridges that gap by introducing an architecture-specific randomized training algorithm that constructs a uniform approximator from $N$ noisy training samples on the $d$-dimensional cube $[0,1]^d$. Our trained neural networks attain the minimax-optimal quantity of \textit{trainable} (non-random) parameters, subject to logarithmic factors which vanish under the idealized noiseless sampling assumed in classical UATs. Additionally, our trained models replicate key behaviours of real-world neural networks, absent in standard UAT constructions, by: (1) exhibiting sub-linear parametric complexity when fine-tuning on structurally related and favourable out-of-distribution tasks, (2) exactly interpolating the training data, and (3) maintaining reasonable Lipschitz regularity (after the initial clustering attention layer). These properties bring state-of-the-art UATs closer to practical machine learning, shifting the central open question from algorithmic implementability with noisy samples to whether stochastic gradient descent can achieve comparable guarantees.

Storm surge is one of the deadliest hazards posed by tropical cyclones (TCs), yet assessing its current and future risk is difficult due to the phenomenon's rarity and physical complexity. Recent advances in artificial intelligence applications to natural hazard modeling suggest a new avenue for addressing this problem. We utilize a deep learning storm surge model to efficiently estimate coastal surge risk in the United States from 900,000 synthetic TC events, accounting for projected changes in TC behavior and sea levels. The derived historical 100-year surge (the event with a 1% yearly exceedance probability) agrees well with historical observations and other modeling techniques. When coupled with an inundation model, we find that heightened TC intensities and sea levels by the end of the century result in a 50% increase in population at risk. Key findings include markedly heightened risk in Florida, and critical thresholds identified in Georgia and South Carolina.

The field of cybersecurity NER lacks standardized labels, making it challenging to combine datasets. We investigate label unification across four cybersecurity datasets to increase data resource usability. We perform a coarse-grained label unification and conduct pairwise cross-dataset evaluations using BiLSTM models. Qualitative analysis of predictions reveals errors, limitations, and dataset differences. To address unification limitations, we propose alternative architectures including a multihead model and a graph-based transfer model. Results show that models trained on unified datasets generalize poorly across datasets. The multihead model with weight sharing provides only marginal improvements over unified training, while our graph-based transfer model built on BERT-base-NER shows no significant performance gains compared BERT-base-NER.

Modern language models are trained on large amounts of data. These data inevitably include controversial and stereotypical content, which contains all sorts of biases related to gender, origin, age, etc. As a result, the models express biased points of view or produce different results based on the assigned personality or the personality of the user. In this paper, we investigate various proxy measures of bias in large language models (LLMs). We find that evaluating models with pre-prompted personae on a multi-subject benchmark (MMLU) leads to negligible and mostly random differences in scores. However, if we reformulate the task and ask a model to grade the user's answer, this shows more significant signs of bias. Finally, if we ask the model for salary negotiation advice, we see pronounced bias in the answers. With the recent trend for LLM assistant memory and personalization, these problems open up from a different angle: modern LLM users do not need to pre-prompt the description of their persona since the model already knows their socio-demographics.

Large Language Models (LLMs) need to adapt their predictions to diverse cultural contexts to benefit diverse communities across the world. While previous efforts have focused on single-LLM, single-turn approaches, we propose to exploit the complementary strengths of multiple LLMs to promote cultural adaptability. We introduce a Multi-Agent Debate framework, where two LLM-based agents debate over a cultural scenario and collaboratively reach a final decision. We propose two variants: one where either LLM agents exclusively debate and another where they dynamically choose between self-reflection and debate during their turns. We evaluate these approaches on 7 open-weight LLMs (and 21 LLM combinations) using the NormAd-ETI benchmark for social etiquette norms in 75 countries. Experiments show that debate improves both overall accuracy and cultural group parity over single-LLM baselines. Notably, multi-agent debate enables relatively small LLMs (7-9B) to achieve accuracies comparable to that of a much larger model (27B parameters).

Natural disasters, including earthquakes, wildfires and cyclones, bear a huge risk on human lives as well as infrastructure assets. An effective response to disaster depends on the ability to rapidly and efficiently assess the intensity of damage. Artificial Intelligence (AI) and Generative Artificial Intelligence (GenAI) presents a breakthrough solution, capable of combining knowledge from multiple types and sources of data, simulating realistic scenarios of disaster, and identifying emerging trends at a speed previously unimaginable. In this paper, we present a comprehensive review on the prospects of AI and GenAI in damage assessment for various natural disasters, highlighting both its strengths and limitations. We talk about its application to multimodal data such as text, image, video, and audio, and also cover major issues of data privacy, security, and ethical use of the technology during crises. The paper also recognizes the threat of Generative AI misuse, in the form of dissemination of misinformation and for adversarial attacks. Finally, we outline avenues of future research, emphasizing the need for secure, reliable, and ethical Generative AI systems for disaster management in general. We believe that this work represents the first comprehensive survey of Gen-AI techniques being used in the field of Disaster Assessment and Response.

Sparse observations and coarse-resolution climate models limit effective regional decision-making, underscoring the need for robust downscaling. However, existing AI methods struggle with generalization across variables and geographies and are constrained by the quadratic complexity of Vision Transformer (ViT) self-attention. We introduce ORBIT-2, a scalable foundation model for global, hyper-resolution climate downscaling. ORBIT-2 incorporates two key innovations: (1) Residual Slim ViT (Reslim), a lightweight architecture with residual learning and Bayesian regularization for efficient, robust prediction; and (2) TILES, a tile-wise sequence scaling algorithm that reduces self-attention complexity from quadratic to linear, enabling long-sequence processing and massive parallelism. ORBIT-2 scales to 10 billion parameters across 65,536 GPUs, achieving up to 4.1 exaFLOPS sustained throughput and 74--98% strong scaling efficiency. It supports downscaling to 0.9 km global resolution and processes sequences up to 4.2 billion tokens. On 7 km resolution benchmarks, ORBIT-2 achieves high accuracy with $R^2$ scores in the range of 0.98--0.99 against observational data.

--The introduction of multimodal models is a huge step forward in Artificial Intelligence. A single model is trained to understand multiple modalities: text, image, video, and audio. Open-source multimodal models have made these breakthroughs more accessible. However, considering the vast landscape of adversarial attacks across these modalities, these models also inherit vulnerabilities of all the modalities, and ultimately, the adversarial threat amplifies. While broad research is available on possible attacks within or across these modalities, a practitioner-focused view that outlines attack types remains absent in the multimodal world. As more Machine Learning Practitioners adopt, fine-tune, and deploy open-source models in real-world applications, it's crucial that they can view the threat landscape and take the preventive actions necessary. This paper addresses the gap by surveying adversarial attacks targeting all four modalities: text, image, video, and audio. This survey provides a view of the adversarial attack landscape and presents how multimodal adversarial threats have evolved. T o the best of our knowledge, this survey is the first comprehensive summarization of the threat landscape in the multimodal world. The advent of models that can comprehend and create content on multiple data types such as Text, Images, Video, and Audio is no less than revolutionary. Multimodal models have shown extremely advanced comprehension and generation abilities. The open-source community has also been a catalyst in developing and deploying such capabilities.

Large language models (LLMs) are increasingly capable of simulating human behavior, offering cost-effective ways to estimate user responses to various surveys and polls. However, the questions in these surveys usually reflect socially understood attitudes: the patterns of attitudes of old/young, liberal/conservative, as understood by both members and non-members of those groups. It is not clear whether the LLM binding is \emph{deep}, meaning the LLM answers as a member of a particular in-group would, or \emph{shallow}, meaning the LLM responds as an out-group member believes an in-group member would. To explore this difference, we use questions that expose known in-group/out-group biases. This level of fidelity is critical for applying LLMs to various political science studies, including timely topics on polarization dynamics, inter-group conflict, and democratic backsliding. To this end, we propose a novel methodology for constructing virtual personas with synthetic user "backstories" generated as extended, multi-turn interview transcripts. This approach is justified by the theory of \emph{narrative identity} which argues that personality at the highest level is \emph{constructed} from self-narratives. Our generated backstories are longer, rich in detail, and consistent in authentically describing a singular individual, compared to previous methods. We show that virtual personas conditioned on our backstories closely replicate human response distributions (up to an 87% improvement as measured by Wasserstein Distance) and produce effect sizes that closely match those observed in the original studies of in-group/out-group biases. Altogether, our work extends the applicability of LLMs beyond estimating socially understood responses, enabling their use in a broader range of human studies.

Large Language and Vision Models using a Mixture-of-Experts (MoE) architecture pose significant challenges for deployment due to their computational and memory demands. Mixed Precision Quantization assigns different precisions to different layers of an LLM/VLM based on layer sensitivity and importance within the model. In this work, we propose a Post Training Quantization algorithm, MoPEQ, that assigns optimal bit width to each expert. Our method balances accuracy and model size by analyzing each expert's sensitivity using Hessian trace approximation instead of relying on the activation frequency of the expert. This per-expert granularity approach clusters similar experts to maintain model performance while reducing memory requirements. The experimental results on VLMEvalKit benchmark datasets using State-of-the-art VLMs Deepseek-VL2 -tiny, -small, -base, and MolmoE models demonstrate that our mixed precision quantized MoEs achieve competitive accuracy with substantial improvements in memory footprint compared to uniform-precision baseline methods. W e perform a comprehensive study to analyze the impact of expert activation frequency and sensitivity using Hessian trace approximation at both layer-wise and model-wide expert precision allocation of 2, 3, and 4 bits to provide a thorough understanding of mixed precision quantization of VLM-MoEs. The code is available here.