搜索结果：Minerva anestesiologica

Recent advancements in video models have shown tremendous progress, particularly in long video understanding. However, current benchmarks predominantly feature western-centric data and English as the dominant language, introducing significant biases in evaluation. To address this, we introduce MINERVA-Cultural, a challenging benchmark for multicultural and multilingual video reasoning. MINERVA-Cultural comprises high-quality, entirely human-generated annotations from diverse, region-specific cultural videos across 18 global locales. Unlike prior work that relies on automatic translations, MINERVA-Cultural provides complex questions, answers, and multi-step reasoning steps, all crafted in native languages. Making progress on MINERVA-Cultural requires a deeply situated understanding of visual cultural context. Furthermore, we leverage MINERVA-Cultural's reasoning traces to construct evidence-based graphs and propose a novel iterative strategy using these graphs to identify fine-grained errors in reasoning. Our evaluations reveal that SoTA Video-LLMs struggle significantly, performing substantially below human-level accuracy, with errors primarily stemming from the visual perception o

Three decades ago, humanity entered the Exoplanet Era, with the discovery of the first planets orbiting other stars. Today, more than 6000 exoplanets are known - a tally recently bolstered by NASA's TESS spacecraft. Whilst TESS is an exceptional planet finding machine, dedicated follow-up observations from the ground are required to confirm the existence of the planets it discovers. To achieve this, we constructed the southern hemisphere's only dedicated exoplanet detection and characterisation facility, MINERVA-Australis, at the University of Southern Queensland's Mt Kent Observatory. Funded in 2015, MINERVA-Australis saw first light in 2018, in time for the launch of TESS. MINERVA-Australis has since been scouring the skies, working to confirm and characterise the incredible harvest of planets detected by TESS. To date, the facility has contributed to the discovery of 40 new exoplanets, and continued the legacy of radial velocity data from the Anglo-Australian Planet Search program.

We compare recent MINERvA antineutrino-hydrogen charged-current measurements to phenomenological predictions of the axial-vector form factor based on fits to all available electron scattering and deuterium bubble-chamber data and to representative lattice-QCD (LQCD) determination by the PNDME Collaboration. While there is $1$--$2σ$ agreement in the cross section with MINERvA data for each bin in $Q^2$, we identify three regions with different relevance and opportunity for LQCD predictions. For $Q^2 \lesssim 0.2~\mathrm{GeV}^2$, the phenomenological extractions have large number of data points and LQCD is competitive, while MINERvA data have large errors. For $0.2~\mathrm{GeV}^2 \lesssim Q^2 \lesssim 1~\mathrm{GeV}^2$, LQCD is competitive with the MINERvA determination, and both give values larger than from phenomenological extraction. For $Q^2 > 1~\mathrm{GeV}^2$, the MINERvA data are the most precise. Our analysis indicates that with improving precision of MINERvA-like experiments and LQCD data, the uncertainty in the nucleon axial-vector form factor will be steadily reduced.

We present an overview of the MINERVA survey, a 259.8 hour (prime) and 127 hour (parallel) Cycle 4 treasury program on the James Webb Space Telescope (JWST). MINERVA is obtaining 8 filter NIRCam medium band imaging (F140M, F162M, F182M, F210M, F250M, F300M, F360M, F460M) and 2 filter MIRI imaging (F1280W, F1500W) in four of the five CANDELS Extragalactic fields: UDS, COSMOS, AEGIS and GOODS-N. These fields were previously observed in Cycle 1 with 7 - 9 NIRCam filters by the PRIMER, CEERS and JADES programs. MINERVA reaches a 5$σ$ depth of 28.1 mag in F300M and covers $\sim$ 542 arcmin$^2$, increasing the area of existing JWST medium-band coverage in at least 8 bands by $\sim$ 7$\times$. The MIRI imaging reaches a 5$σ$ depth of 23.9 mag in F1280W and covers $\sim$ 275 arcmin$^2$ in at least 2 MIRI filters. When combined with existing imaging, these data will provide a photometric catalog with 20-26 JWST filters (depending on field) and 26-35 filters total, including HST. This paper presents a detailed breakdown of the filter coverage, exposure times, and field layout relative to previous observations, as well as an overview of the primary science goals of the project. These include

Despite advances in AI for contact centers, customer experience (CX) continues to suffer from high average handling time (AHT), low first-call resolution, and poor customer satisfaction (CSAT). A key driver is the cognitive load on agents, who must navigate fragmented systems, troubleshoot manually, and frequently place customers on hold. Existing AI-powered agent-assist tools are often reactive driven by static rules, simple prompting, or retrieval-augmented generation (RAG) without deeper contextual reasoning. We introduce Agentic AI goal-driven, autonomous, tool-using systems that proactively support agents in real time. Unlike conventional approaches, Agentic AI identifies customer intent, triggers modular workflows, maintains evolving context, and adapts dynamically to conversation state. This paper presents a case study of Minerva CQ, a real-time Agent Assist product deployed in voice-based customer support. Minerva CQ integrates real-time transcription, intent and sentiment detection, entity recognition, contextual retrieval, dynamic customer profiling, and partial conversational summaries enabling proactive workflows and continuous context-building. Deployed in live product

Multimodal LLMs are turning their focus to video benchmarks, however most video benchmarks only provide outcome supervision, with no intermediate or interpretable reasoning steps. This makes it challenging to assess if models are truly able to combine perceptual and temporal information to reason about videos, or simply get the correct answer by chance or by exploiting linguistic biases. To remedy this, we provide a new video reasoning dataset called MINERVA for modern multimodal models. Each question in the dataset comes with 5 answer choices, as well as detailed, hand-crafted reasoning traces. Our dataset is multimodal, diverse in terms of video domain and length, and consists of complex multi-step questions. Extensive benchmarking shows that our dataset provides a challenge for frontier open-source and proprietary models. We perform fine-grained error analysis to identify common failure modes across various models, and create a taxonomy of reasoning errors. We use this to explore both human and LLM-as-a-judge methods for scoring video reasoning traces, and find that failure modes are primarily related to temporal localization, followed by visual perception errors, as opposed to