Retrieval-Augmented Generation (RAG) is a prevalent approach to infuse a private knowledge base of documents with Large Language Models (LLM) to build Generative Q\&A (Question-Answering) systems. However, RAG accuracy becomes increasingly challenging as the corpus of documents scales up, with Retrievers playing an outsized role in the overall RAG accuracy by extracting the most relevant document from the corpus to provide context to the LLM. In this paper, we propose the 'Blended RAG' method of leveraging semantic search techniques, such as Dense Vector indexes and Sparse Encoder indexes, blended with hybrid query strategies. Our study achieves better retrieval results and sets new benchmarks for IR (Information Retrieval) datasets like NQ and TREC-COVID datasets. We further extend such a 'Blended Retriever' to the RAG system to demonstrate far superior results on Generative Q\&A datasets like SQUAD, even surpassing fine-tuning performance.

This paper presents a pilot study aimed at introducing multi-agent debate into multimodal reasoning. The study addresses two key challenges: the trivialization of opinions resulting from excessive summarization and the diversion of focus caused by distractor concepts introduced from images. These challenges stem from the inductive (bottom-up) nature of existing debating schemes. To address the issue, we propose a deductive (top-down) debating approach called Blueprint Debate on Graphs (BDoG). In BDoG, debates are confined to a blueprint graph to prevent opinion trivialization through world-level summarization. Moreover, by storing evidence in branches within the graph, BDoG mitigates distractions caused by frequent but irrelevant concepts. Extensive experiments validate BDoG, achieving state-of-the-art results in Science QA and MMBench with significant improvements over previous methods.

We present a new method to automatically generate semantic segmentation annotations for thermal imagery captured from an aerial vehicle by utilizing satellite-derived data products alongside onboard global positioning and attitude estimates. This new capability overcomes the challenge of developing thermal semantic perception algorithms for field robots due to the lack of annotated thermal field datasets and the time and costs of manual annotation, enabling precise and rapid annotation of thermal data from field collection efforts at a massively-parallelizable scale. By incorporating a thermal-conditioned refinement step with visual foundation models, our approach can produce highly-precise semantic segmentation labels using low-resolution satellite land cover data for little-to-no cost. It achieves 98.5% of the performance from using costly high-resolution options and demonstrates between 70-160% improvement over popular zero-shot semantic segmentation methods based on large vision-language models currently used for generating annotations for RGB imagery. Code will be available at: https://github.com/connorlee77/aerial-auto-segment.

The robot uses its chest, hips and arms to handle objects -- just like we do. Theo works weekdays, weekends and nights and never complains about a sore spine despite performing hour upon hour of what, for a regular farm hand, would be backbreaking labor checking Dutch tulip fields for sick flowers. The boxy robot -- named after a retired employee at the WAM Pennings farm near the Dutch North Sea coast -- is a new high-tech weapon in the battle to root out disease from the bulb fields as they erupt into a riot of springtime color. On a windy spring morning, the robot trundled Tuesday along rows of yellow and red "goudstuk" tulips, checking each plant and, when necessary, killing diseased bulbs to prevent the spread of the tulip-breaking virus. The dead bulbs are removed from healthy ones in a sorting warehouse after they have been harvested.

This study investigates lightning at tall objects and evaluates the risk of upward lightning (UL) over the eastern Alps and its surrounding areas. While uncommon, UL poses a threat, especially to wind turbines, as the long-duration current of UL can cause significant damage. Current risk assessment methods overlook the impact of meteorological conditions, potentially underestimating UL risks. Therefore, this study employs random forests, a machine learning technique, to analyze the relationship between UL measured at Gaisberg Tower (Austria) and $35$ larger-scale meteorological variables. Of these, the larger-scale upward velocity, wind speed and direction at 10 meters and cloud physics variables contribute most information. The random forests predict the risk of UL across the study area at a 1 km$^2$ resolution. Strong near-surface winds combined with upward deflection by elevated terrain increase UL risk. The diurnal cycle of the UL risk as well as high-risk areas shift seasonally. They are concentrated north/northeast of the Alps in winter due to prevailing northerly winds, and expanding southward, impacting northern Italy in the transitional and summer months. The model performs best in winter, with the highest predicted UL risk coinciding with observed peaks in measured lightning at tall objects. The highest concentration is north of the Alps, where most wind turbines are located, leading to an increase in overall lightning activity. Comprehensive meteorological information is essential for UL risk assessment, as lightning densities are a poor indicator of lightning at tall objects.

From The Terminator to The Day After Tomorrow, movies have envisioned just about every possibility for how the world might end. If you're a science fiction movie buff, you might think that some of these apocalyptic scenarios seem a little far-fetched. But hold onto your popcorn, as experts say that some of these disastrous plotlines could actually become a reality. While we don't need to worry about an asteroid wiping us out like in Armageddon, experts warn that a bioweapon leak like 12 Monkeys could really end the world. And if your favourite blockbuster does give us a glimpse at how the world will end, not even Bruce Willis will be able to save us. Apocalypse movies find their inspiration in a number of different disasters, but which are the most realistic. An escaped bioweapon could pose a genuine threat of destroying humanity.

In recent years, atmospheric research has received increasing attention from environmental experts and the public because atmospheric phenomena such as El Nino, global warming, ozone depletion, and drought that may have negative effects on the Earth's climate and ecosystem are occurring more often (Walther et al. 2002; Karl and Trenberth 2003; Trenberth et al. 2014). In order to model the status quo and predict the trend of atmospheric phenomena and events, researchers need to retrieve data from various relevant domains, such as chemical components of aerosols and gases, the terrestrial surface, energy consumption, the hydrosphere, the biosphere, etc. (Schneider, 2006; Fowler et al., 2009; Guilyardi et al, 2009; Ramanathan et al., 2011; Katul et al., 2012). In complex earth system modeling, the data and services for atmospheric study present the characteristics of being distributed, collaborative and adaptive (Plale et al., 2006). The massive volume, rapid velocity and wide variety of data has led to a new era of atmospheric research that consists of accessing and integrating big data from distributed sources, conducting collaborative analysis in an interactive way, providing intelligent services for data management, and integration and visualization to foster discovery of hidden or new knowledge. One of the most important ways to support these activities is to establish a national or international spatial data infrastructure and geospatial cyberinfrastructure on which the data and computational resources can be easily shared, the spatial analysis tool can be executed on-the-fly and the scientific results can be effectively visualized (Yang et al., 2008; Li et al., 2011). Technically, a geospatial cyberinfrastructure (GCI) is an architecture that effectively utilizes geo-referenced data to connect people, information and computers based on the standardized data access protocols, high speed internet, high-performance computing facilities (HPC) and service-oriented data management (Yang et al., 2010). Since the concept's official introduction by the National Science Foundation (NSF) in its 2003 blue ribbon report, cyberinfrastructure research has attracted much attention from the atmospheric science domain because of its promise of bringing paradigm change for

ChatGPT caught regulators by surprise when it set off a new AI race. As companies have rushed to develop and release ever more powerful models, lawmakers and regulators around the world have sought to catch up and rein in development. As governments spin up new AI programs, regulators around the world are urgently trying to hire AI experts. But some of the job ads are raising eyebrows and even chuckles among AI researchers and engineers for offering wages that, amid the current AI boom, look pitiful. The European AI Office, which will be central to the implementation of the EU's AI Act, listed vacancies early this month and wants applicants to begin work in the fall.

The recently introduced continuous Skip-gram model is an efficient method for learning high-quality distributed vector representations that capture a large number of precise syntactic and semantic word relationships. In this paper we present several extensions that improve both the quality of the vectors and the training speed. By subsampling of the frequent words we obtain significant speedup and also learn more regular word representations. We also describe a simple alternative to the hierarchical softmax called negative sampling. An inherent limitation of word representations is their indifference to word order and their inability to represent idiomatic phrases. For example, the meanings of "Canada" and "Air" cannot be easily combined to obtain "Air Canada". Motivated by this example, we present a simple method for finding phrases in text, and show that learning good vector representations for millions of phrases is possible.

Accurate and efficient analysis of multivariate spatio-temporal data is critical in climatology, geology, and sociology applications. Existing models usually assume simple inter-dependence among variables, space, and time, and are computationally expensive. We propose a unified low rank tensor learning framework for multivariate spatio-temporal analysis, which can conveniently incorporate different properties in spatio-temporal data, such as spatial clustering and shared structure among variables. We demonstrate how the general framework can be applied to cokriging and forecasting tasks, and develop an efficient greedy algorithm to solve the resulting optimization problem with convergence guarantee. We conduct experiments on both synthetic datasets and real application datasets to demonstrate that our method is not only significantly faster than existing methods but also achieves lower estimation error.