arXiv subjects

Roger H. French

Publications and source records attributed to Roger H. French.

3 recordsLinked to original sources

arXiv2026

Context Determines Optimal Architecture in Materials Segmentation

Segmentation architectures are typically benchmarked on single imaging modalities, obscuring deployment-relevant performance variations: an architecture optimal for one modality may underperform on another. We present a cross-modal evaluation framework for materials image segmentation spanning SEM, AFM, XCT, and optical microscopy. Our evaluation of six encoder-decoder combinations across seven datasets reveals that optimal architectures vary systematically by context: UNet excels for high-contrast 2D imaging while DeepLabv3+ is preferred for the hardest cases. The framework also provides deployment feedback via out-of-distribution detection and counterfactual explanations that reveal which microstructural features drive predictions. Together, the architecture guidance, reliability signals, and interpretability tools address a practical gap in materials characterization, where researchers lack tools to select architectures for their specific imaging setup or assess when models can be trusted on new samples.

cs.CV↗

arXiv2025

GeoOutageKG: A Multimodal Geospatiotemporal Knowledge Graph for Multiresolution Power Outage Analysis

Detecting, analyzing, and predicting power outages is crucial for grid risk assessment and disaster mitigation. Numerous outages occur each year, exacerbated by extreme weather events such as hurricanes. Existing outage data are typically reported at the county level, limiting their spatial resolution and making it difficult to capture localized patterns. However, it offers excellent temporal granularity. In contrast, nighttime light satellite image data provides significantly higher spatial resolution and enables a more comprehensive spatial depiction of outages, enhancing the accuracy of assessing the geographic extent and severity of power loss after disaster events. However, these satellite data are only available on a daily basis. Integrating spatiotemporal visual and time-series data sources into a unified knowledge representation can substantially improve power outage detection, analysis, and predictive reasoning. In this paper, we propose GeoOutageKG, a multimodal knowledge graph that integrates diverse data sources, including nighttime light satellite image data, high-resolution spatiotemporal power outage maps, and county-level timeseries outage reports in the U.S. We describe our method for constructing GeoOutageKG by aligning source data with a developed ontology, GeoOutageOnto. Currently, GeoOutageKG includes over 10.6 million individual outage records spanning from 2014 to 2024, 300,000 NTL images spanning from 2012 to 2024, and 15,000 outage maps. GeoOutageKG is a novel, modular and reusable semantic resource that enables robust multimodal data integration. We demonstrate its use through multiresolution analysis of geospatiotemporal power outages.

cs.IR↗

arXiv2012

Chirality-Dependent Properties of Carbon Nanotubes. Electronic Structure: Optical Dispersion Properties, Hamaker Coefficients and van der Waals - London dispersion interactions

Optical dispersion spectra at energies up to 30 eV play a vital role in understanding the chirality-dependent van der Waals London dispersion interactions of single wall carbon nanotubes (SWCNTs). We use one-electron theory based calculations to obtain the band structures and the frequency dependent dielectric response function from 0-30 eV for 64 SWCNTs differing in radius, electronic structure classification, and geometry. The resulting optical dispersion properties can be categorized over three distinct energy intervals (M, pi, and sigma, respectively representing 0-0.1, 0.1-5, and 5-30 eV regions) and over radii above or below the zone-folding limit of 0.7 nm. While pi peaks vary systematically with radius for a given electronic structure type, peaks are independent of tube radius above the zone folding limit and depend entirely on SWCNT geometry. Based on these calculated one-electron dielectric response functions we compute and review Van derWaals - London dispersion spectra, full spectral Hamaker coefficients, and van derWaals - London dispersion interaction energies for all calculated frequency dependent dielectric response functions. Our results are categorized using a new optical dielectric function classification scheme that groups the nanotubes according to observable trends and notable features (e.g. the metal paradox) in the 0-30 eV part of the optical dispersion spectra. While the trends in these spectra begin to break down at the zone folding diameter limit, the trends in the related van derWaals - London dispersion spectra tend to remain stable all the way down to the smallest single wall carbon nanotubes in a given class.

cond-mat.mes-hall↗