Search

This paper introduces Moonshine, a family of speech recognition models optimized for live transcription and voice command processing. Moonshine is based on an encoder-decoder transformer architecture and employs Rotary Position Embedding (RoPE) instead of traditional absolute position embeddings. The model is trained on speech segments of various lengths, but without using zero-padding, leading to greater efficiency for the encoder during inference time. When benchmarked against OpenAI's Whisper tiny-en, Moonshine Tiny demonstrates a 5x reduction in compute requirements for transcribing a 10-second speech segment while incurring no increase in word error rates across standard evaluation datasets. These results highlight Moonshine's potential for real-time and resource-constrained applications., Comment: 7 pages, 6 figures, 3 tables

Super-resolution ultrasound (SRUS) visualises microvasculature beyond the ultrasound diffraction limit (wavelength($\lambda$)/2) by localising and tracking spatially isolated microbubble contrast agents. SRUS phantoms typically consist of simple tube structures, where diameter channels below 100 $\mu$m are not available. Furthermore, these phantoms are generally fragile and unstable, have limited ground truth validation, and their simple structure limits the evaluation of SRUS algorithms. To aid SRUS development, robust and durable phantoms with known and physiologically relevant microvasculature are needed for repeatable SRUS testing. This work proposes a method to fabricate durable microvascular phantoms that allow optical gauging for SRUS validation. The methodology used a microvasculature negative print embedded in a Polydimethylsiloxane to fabricate a microvascular phantom. Branching microvascular phantoms with variable microvascular density were demonstrated with optically validated vessel diameters down to $\sim$ 60 $\mu$m ($\lambda$/5.8; $\lambda$ =$\sim$ 350 $\mu$m). SRUS imaging was performed and validated with optical measurements. The average SRUS error was 15.61 $\mu$m ($\lambda$/22) with a standard deviation error of 11.44 $\mu$m. The average error decreased to 7.93 $\mu$m ($\lambda$/44) once the number of localised microbubbles surpassed 1000 per estimated diameter. In addition, the less than 10$\%$ variance of acoustic and optical properties and the mechanical toughness of the phantoms measured a year after fabrication demonstrated their long-term durability. This work presents a method to fabricate durable and optically validated complex microvascular phantoms which can be used to quantify SRUS performance and facilitate its further development., Comment: This work has been submitted to the IEEE for possible publication

We present an Artificial Neural Network (ANN) model of photospheric lithium depletion in cool stars (3000 < Teff / K < 6500), producing estimates and probability distributions of age from Li I 6708A equivalent width (LiEW) and effective temperature data inputs. The model is trained on the same sample of 6200 stars from 52 open clusters, observed in the Gaia-ESO spectroscopic survey, and used to calibrate the previously published analytical EAGLES model, with ages 2 - 6000 Myr and -0.3 < [Fe/H] < 0.2. The additional flexibility of the ANN provides some improvements, including better modelling of the "lithium dip" at ages < 50 Myr and Teff ~ 3500K, and of the intrinsic dispersion in LiEW at all ages. Poor age discrimination is still an issue at ages > 1 Gyr, confirming that additional modelling flexibility is not sufficient to fully represent the LiEW - age - Teff relationship, and suggesting the involvement of further astrophysical parameters. Expansion to include such parameters - rotation, accretion, and surface gravity - is discussed, and the use of an ANN means these can be more easily included in future iterations, alongside more flexible functional forms for the LiEW dispersion. Our methods and ANN model are provided in an updated version 2.0 of the EAGLES software., Comment: Accepted for publication in Monthly Notices of the Royal Astronomical Society. Code available at https://github.com/robdjeff/eagles. Electronic tables are available from the author

Formal mentoring has been an effective approach to mitigate challenges facing underserved youth and contributes to observable improvements in behavior, relationships, and emotional well-being. In recent years, virtual mentoring has emerged as a promising way to expand the provision of mentoring. By engaging virtually, mentorship programs can mitigate typical challenges of travel time and distance between mentor and mentee. Nonetheless, there are also challenges associated with virtual mentoring, including the potential for miscommunication and possible difficulties developing deep connections. In 2020, the U.S. Dream Academy began developing and implementing virtual and hybrid mentoring services to explore the benefits and address the challenges of virtual mentoring. As the coronavirus disease 2019 (COVID-19) pandemic caused widespread disruption in schooling in the United States, the U.S. Dream Academy leveraged this unique opportunity to begin implementing virtual mentoring. This report describes how the U.S. Dream Academy sought to expand its mentoring program to include a virtual option. The authors summarize the development and implementation of the mentoring program and lessons learned from this experience with the goal of supporting program improvement. This report may also assist other mentoring programs that seek to adopt a similar mentoring model. [The research for this report was sponsored by U.S. Dream Academy.]

Formal mentoring has been an effective approach to mitigate challenges facing underserved youth and contributes to observable improvements in behavior, relationships, and emotional well-being (Rhodes, 2008). Mentored youth appear to gain improvements in peer and parent relationships and school performance, while also engaging in lower levels of substance and alcohol use relative to youth who are not mentored (Raposa et al., 2019). Studies have found mentoring effective in preventing psychosocial problems, such as delinquent behavior (DuBois and Karcher, 2013). Despite the evidence on the benefits of formal mentoring, an estimated 9 million young people are growing up in under-resourced environments without the support of formal mentors (Bruce and Bridgeland, 2014). In recent years, virtual mentoring has emerged as a promising way to expand the provision of mentoring (Andersen and Wellen, 2023). By engaging virtually, mentorship programs can mitigate the typical challenges of travel time and distance between mentor and mentee (Ensher, Heun, and Blanchard, 2003) and may even facilitate the development of rapport and trust (Ensher and Murphy, 1997; Miller and Griffiths, 2005). In 2020, the U.S. Dream Academy (hereafter referred to as "Dream"), a longtime innovator in developing out-of-school-time programs and formal mentoring for young people, expanded its mentoring program to include a virtual option in six Learning Centers across the country. RAND researchers collaborated with Dream to evaluate the development and implementation of the mentoring program and the supplemental services offered. This brief illustrates how flexible programming, including the use of in-person, hybrid, and virtual mentoring approaches, allowed for the adaptation of supports to meet the specific needs of youth and families in their communities.

We report a comprehensive study of Sperrylite (PtAs2), the main platinum source in natural minerals, as a function of applied pressures up to 150 GPa. While no structural phase transition was detected from pressure-dependent X-ray measurements, the unit cell volume shrinks monotonically with pressure following the third-order Birch-Murnaghan equation of state. The mildly semiconducting behavior found in pure synthesized crystals at ambient pressures becomes more insulating upon increasing applied pressure before metalizing at higher pressures, giving way to the appearance of an abrupt decrease in resistance near 3 K at pressures above 92 GPa consistent with the onset of a superconducing phase. The pressure evolution of the calculated electronic band structure reveals the same physical trend as our transport measurements, with a non-monotonic evolution explained by a hole band that is pushed below the Fermi energy and an electron band that approaches it as a function of pressure, both reaching a touching point suggestive of an excitonic state. A topological Lifshitz transition of the electronic structure and an increase in the density of states may naturally explain the onset of superconductivity in this material, Comment: 7 pages, 7 figures

Federated Learning (FL) enables local devices to collaboratively learn a shared predictive model by only periodically sharing model parameters with a central aggregator. However, FL can be disadvantaged by statistical heterogeneity produced by the diversity in each local devices data distribution, which creates different levels of Independent and Identically Distributed (IID) data. Furthermore, this can be more complex when optimising different combinations of FL parameters and choosing optimal aggregation. In this paper, we present an empirical analysis of different FL training parameters and aggregators over various levels of statistical heterogeneity on three datasets. We propose a systematic data partition strategy to simulate different levels of statistical heterogeneity and a metric to measure the level of IID. Additionally, we empirically identify the best FL model and key parameters for datasets of different characteristics. On the basis of these, we present recommended guidelines for FL parameters and aggregators to optimise model performance under different levels of IID and with different datasets, Comment: 27 pages, 17 figures

Tiny machine learning (TinyML) for low-power devices lacks robust datasets for development. We present Wake Vision, a large-scale dataset for person detection that contains over 6 million quality-filtered images. We provide two variants: Wake Vision (Large) and Wake Vision (Quality), leveraging the large variant for pretraining and knowledge distillation, while the higher-quality labels drive final model performance. The manually labeled validation and test sets reduce error rates from 7.8% to 2.2% compared to previous standards. In addition, we introduce five detailed benchmark sets to evaluate model performance in real-world scenarios, including varying lighting, camera distances, and demographic characteristics. Training with Wake Vision improves accuracy by 1.93% over existing datasets, demonstrating the importance of dataset quality for low-capacity models and dataset size for high-capacity models. The dataset, benchmarks, code, and models are available under the CC-BY 4.0 license, maintained by the Edge AI Foundation.

We prove a differential version of the Artin-Rees lemma with the use of Noetherian differential operators. As a consequence, we obtain several uniformity results for nonreduced rings.

Bernstein's inequality is a central result in the theory of $D$-modules on smooth varieties. While Bernstein's inequality fails for rings of differential operators on general singularities, recent work of \`{A}lvarez Montaner, Hern\'andez, Jeffries, N\'u\~nez-Betancourt, Teixeira, and Witt establishes Bernstein's inequality for invariants of finite groups in characteristic zero and certain other mild singularities in positive characteristic. Motivated by extending this result to new classes of singular rings, we introduce a ``two-sided'' analogue of the Bernstein-Sato polynomial which we call the sandwich Bernstein-Sato polynomial. We apply this notion to give an effective criterion to verify Bernstein's inequality, and apply this to show that Bernstein's inequality holds for the coordinate ring of $\mathbb{P}^a \times \mathbb{P}^b$ via the Segre embedding. We also establish a number of examples and basic results on sandwich Bernstein-Sato polynomials.

High-entropy alloys have drawn recent interest for their promising mechanical properties and irradiation resistance. Various properties, namely transport properties, are controlled by point defect concentration, which must be known before performing atomistic simulations to compute transport coefficients. In this work, we present a general Langmuir-like model for impurity concentration in an arbitrarily complex solid solution and apply this model to generate expressions for concentrations of vacancies and small interstitial atoms. We then calculate the vacancy concentration as a function of temperature in the equiatomic CoNiCrFeMn and FeAl alloys with modified embedded-atom-method potentials for various chemical orderings, showing there is no clear correlation between vacancy thermodynamics and chemical ordering in the CoNiCrFeMn alloy but clear systematic patterns for FeAl., Comment: 13 pages, 6 figures

The Gaia-ESO Survey is an European Southern Observatory (ESO) public spectroscopic survey that targeted $10^5$ stars in the Milky Way covering the major populations of the disk, bulge and halo. The observations were made using FLAMES on the VLT obtaining both UVES high ($R\sim47,000$) and GIRAFFE medium ($R\sim20,000$) resolution spectra. The analysis of the Gaia-ESO spectra was the work of multiple analysis teams (nodes) within five working groups (WG). The homogenisation of the stellar parameters within WG11 (high resolution observations of FGK stars) and the homogenisation of the stellar parameters within WG10 (medium resolution observations of FGK stars) is described here. In both cases, the homogenisation was carried out using a bayesian Inference method developed specifically for the Gaia-ESO Survey by WG11. The WG10 homogenisation primarily used the cross-match of stars with WG11 as the reference set in both the stellar parameter and chemical abundance homogenisation. In this way the WG10 homogenised results have been placed directly onto the WG11 stellar parameter and chemical abundance scales. The reference set for the metal-poor end was sparse which limited the effectiveness of the homogenisation in that regime. For WG11, the total number of stars for which stellar parameters were derived was 6,231 with typical uncertainties for Teff, log g and [Fe/H] of 32~K, 0.05 and 0.05 respectively. One or more chemical abundances out of a possible 39 elements were derived for 6,188 of the stars. For WG10, the total number of stars for which stellar parameters were derived was 76,675 with typical uncertainties for Teff, log g and [Fe/H] of 64~K, 0.15 and 0.07 respectively. One or more chemical abundances out of a possible 30 elements were derived for 64,177 of the stars., Comment: 31 pages, 19 figures

The High-Resolution Multi-Object Spectrograph (HRMOS) is a facility instrument that we plan to propose for the Very Large Telescope (VLT) of the European Southern Observatory (ESO), following the initial presentation at the VLT 2030 workshop held at ESO in June 2019. HRMOS provides a combination of capabilities that are essential to carry out breakthrough science across a broad range of active research areas from stellar astrophysics and exoplanet studies to Galactic and Local Group archaeology. HRMOS fills a gap in capabilities amongst the landscape of future instrumentation planned for the next decade. The key characteristics of HRMOS will be high spectral resolution (R = 60000 - 80000) combined with multi-object (20-100) capabilities and long term stability that will provide excellent radial velocity precision and accuracy (10m/s). Initial designs predict that a SNR~100 will be achievable in about one hour for a star with mag(AB) = 15, while with the same exposure time a SNR~ 30 will be reached for a star with mag(AB) = 17. The combination of high resolution and multiplexing with wavelength coverage extending to relatively blue wavelengths (down to 380\,nm), makes HRMOS a spectrograph that will push the boundaries of our knowledge and that is envisioned as a workhorse instrument in the future. The science cases presented in this White Paper include topics and ideas developed by the Core Science Team with the contributions from the astronomical community, also through the wide participation in the first HRMOS Workshop (https://indico.ict.inaf.it/event/1547/) that took place in Firenze (Italy) in October 2021., Comment: 88 pages, 39 figures. Comments and expressions of interest are welcome by contacting members of the Core Science Team

We introduce $p$-derivations and give a few basic ways in which they act like derivatives by numbers.

We present the first large-scale 3D kinematic study of ~2000 spectroscopically-confirmed young stars (<20 Myr) in 18 star clusters and OB associations (hereafter groups) from the combination of Gaia astrometry and Gaia-ESO Survey spectroscopy. We measure 3D velocity dispersions for all groups, which range from 0.61 to 7.4 km/s (1D velocity dispersions of 0.35 to 4.3 km/s). We find the majority of groups have anisotropic velocity dispersions, suggesting they are not dynamically relaxed. From the 3D velocity dispersions, measured radii and estimates of total mass we estimate the virial state and find that all systems are super-virial when only the stellar mass is considered, but that some systems are sub-virial when the mass of the molecular cloud is taken into account. We observe an approximately linear correlation between the 3D velocity dispersion and the group mass, which would imply that the virial state of groups scales as the square root of the group mass. However, we do not observe a strong correlation between virial state and group mass. In agreement with their virial state we find that nearly all of the groups studied are in the process of expanding and that the expansion is anisotropic, implying that groups were not spherical prior to expansion. One group, Rho Oph, is found to be contracting and in a sub-virial state (when the mass of the surrounding molecular cloud is considered). This work provides a glimpse of the potential of the combination of Gaia and data from the next generation of spectroscopic surveys., Comment: 26 pages, 15 figures, accepted for publication in MNRAS