arXiv subjects

David Robinson

Publications and source records attributed to David Robinson.

At least 19 recordsLinked to original sources

arXiv2026

Multi-layer attentive probing improves transfer of audio representations for bioacoustics

Probing heads map the representations learned from audio by a machine learning model to downstream task labels and are a key component in evaluating representation learning. Most bioacoustic benchmarks use a fixed, low-capacity probe, such as a linear layer on the final encoder layer. While this standardization enables model comparisons, it may bias results by overlooking the interaction between encoder features and probe design. In this work, we systematically study different probing strategies across two bioacoustic benchmarks, BEANs and BirdSet. We evaluate last- and multi-layer probing, across linear and attention probes. We show that larger probe heads that leverage time information have superior performance. Our results suggest that current benchmarks may misrepresent encoder quality when relying on a last-layer probing setup. Multi-layer probing improves downstream task performance across all tested models, while attention probing has superior performance to linear probing for transformer models.

cs.SD↗

arXiv2026

Beyond the Baseband: Adaptive Multi-Band Encoding for Full-Spectrum Bioacoustics Classification

Animals hear and vocalize across frequency ranges that differ substantially from humans, often extending into the ultrasonic domain. Yet most computational bioacoustics systems rely on audio models pre-trained at 16 kHz, restricting their usable bandwidth to the 0-8 kHz baseband and discarding higher-frequency information present in many bioacoustic recordings. We investigate a multi-band encoding framework that decomposes the full spectrum of animal calls into band features and fuses them into a unified representation. Similarity analyses on models show that certain encoders produce decorrelated band embeddings that improve class separation after fusion. Classification experiments on three bioacoustic datasets using eight pre-trained models and five fusion strategies show that fused representations consistently outperform the baseband and time-expansion baselines on two datasets, showing the potential of multi-band methods for full-spectrum encoding of animal calls.

cs.LG↗

arXiv2026

BAGEL: Benchmarking Animal Knowledge Expertise in Language Models

Large language models have shown strong performance on broad-domain knowledge and reasoning benchmarks, but it remains unclear how well language models handle specialized animal-related knowledge under a unified closed-book evaluation protocol. We introduce BAGEL, a benchmark for evaluating animal knowledge expertise in language models. BAGEL is constructed from diverse scientific and reference sources, including bioRxiv, Global Biotic Interactions, Xeno-canto, and Wikipedia, using a combination of curated examples and automatically generated closed-book question-answer pairs. The benchmark covers multiple aspects of animal knowledge, including taxonomy, morphology, habitat, behavior, vocalization, geographic distribution, and species interactions. By focusing on closed-book evaluation, BAGEL measures animal-related knowledge of models without external retrieval at inference time. BAGEL further supports fine-grained analysis across source domains, taxonomic groups, and knowledge categories, enabling a more precise characterization of model strengths and systematic failure modes. Our benchmark provides a new testbed for studying domain-specific knowledge generalization in language models and for improving their reliability in biodiversity-related applications.

cs.CL↗

arXiv2026

Enhancing Box and Block Test with Computer Vision for Post-Stroke Upper Extremity Motor Evaluation

Standard clinical assessments of upper-extremity motor function after stroke either rely on ordinal scoring, which lacks sensitivity, or time-based task metrics, which do not capture movement quality. In this work, we present a computer vision-based framework for analysis of upper-extremity movement during the Box and Block Test (BBT) through world-aligned joint angles of fingers, arm, and trunk without depth sensors or calibration objects. We apply this framework to a dataset of 136 BBT recordings collected from 48 healthy individuals and 7 individuals post stroke. Using unsupervised dimensionality reduction of joint-angle features, we analyze movement patterns without relying on expert clinical labels. The resulting embeddings show separation between healthy movement patterns and stroke-related movement deviations. Importantly, some patients with the same BBT scores can be separated with different postural patterns. These results show that world-aligned joint angles can capture meaningful information of upper-extremity functions beyond standard time-based BBT scores, with no effort from the clinician other than monocular video recordings of the patient using a phone or camera. This work highlights the potential of a camera-based, calibration-free framework to measure movement quality in clinical assessments without changing the widely adopted clinical routine.

cs.CV↗

arXiv2026

Audio-to-Image Bird Species Retrieval without Audio-Image Pairs via Text Distillation

Audio-to-image retrieval offers an interpretable alternative to audio-only classification for bioacoustic species recognition, but learning aligned audio-image representations is challenging due to the scarcity of paired audio-image data. We propose a simple and data-efficient approach that enables audio-to-image retrieval without any audio-image supervision. Our proposed method uses text as a semantic intermediary: we distill the text embedding space of a pretrained image-text model (BioCLIP-2), which encodes rich visual and taxonomic structure, into a pretrained audio-text model (BioLingual) by fine-tuning its audio encoder with a contrastive objective. This distillation transfers visually grounded semantics into the audio representation, inducing emergent alignment between audio and image embeddings without using images during training. We evaluate the resulting model on multiple bioacoustic benchmarks. The distilled audio encoder preserves audio discriminative power while substantially improving audio-text alignment on focal recordings and soundscape datasets. Most importantly, on the SSW60 benchmark, the proposed approach achieves strong audio-to-image retrieval performance exceeding baselines based on zero-shot model combinations or learned mappings between text embeddings, despite not training on paired audio-image data. These results demonstrate that indirect semantic transfer through text is sufficient to induce meaningful audio-image alignment, providing a practical solution for visually grounded species recognition in data-scarce bioacoustic settings.

cs.SD↗

arXiv2026

Compact Hypercube Embeddings for Fast Text-based Wildlife Observation Retrieval

Large-scale biodiversity monitoring platforms increasingly rely on multimodal wildlife observations. While recent foundation models enable rich semantic representations across vision, audio, and language, retrieving relevant observations from massive archives remains challenging due to the computational cost of high-dimensional similarity search. In this work, we introduce compact hypercube embeddings for fast text-based wildlife observation retrieval, a framework that enables efficient text-based search over large-scale wildlife image and audio databases using compact binary representations. Building on the cross-view code alignment hashing framework, we extend lightweight hashing beyond a single-modality setup to align natural language descriptions with visual or acoustic observations in a shared Hamming space. Our approach leverages pretrained wildlife foundation models, including BioCLIP and BioLingual, and adapts them efficiently for hashing using parameter-efficient fine-tuning. We evaluate our method on large-scale benchmarks, including iNaturalist2024 for text-to-image retrieval and iNatSounds2024 for text-to-audio retrieval, as well as multiple soundscape datasets to assess robustness under domain shift. Results show that retrieval using discrete hypercube embeddings achieves competitive, and in several cases superior, performance compared to continuous embeddings, while drastically reducing memory and search cost. Moreover, we observe that the hashing objective consistently improves the underlying encoder representations, leading to stronger retrieval and zero-shot generalization. These results demonstrate that binary, language-based retrieval enables scalable and efficient search over large wildlife archives for biodiversity monitoring systems.

cs.IR↗

arXiv2025

Learning Soil Physics from Partial Knowledge and Data: Partitioning Capillary and Adsorbed Soil Water

Soil physics models have long relied on simplifying assumptions to represent complex processes, yet such assumptions can strongly bias model predictions. Here, we propose a paradigm-shifting differentiable hybrid modeling (DHM) framework that instead of simplifying the unknown, learns it from data. As a proof of concept, we apply the hybrid approach to the challenge of partitioning the soil water retention curve (SWRC) into capillary and adsorbed water components, a problem where traditional assumptions have led to divergent results. The hybrid framework derives this partitioning directly from data while remaining guided by a few parsimonious and universally accepted physical constraints. Using basic soil physical properties as inputs, the hybrid model couples an analytical formula for the dry end of the SWRC with data-driven physics-informed neural networks that learn the wet end, the transition between the two ends, and key soil-specific parameters. The model was trained on a SWRC dataset from 482 undisturbed soil samples from Central Europe, spanning a broad range of soil texture classes and organic carbon contents. The hybrid model successfully learned both the overall shape and the capillary and adsorbed components of the SWRC. Notably, the model revealed physically meaningful pore-scale features without relying on explicit geometrical assumptions about soil pore shape or its distribution. Moreover, the model revealed a distinctly nonlinear transition between capillary and adsorbed domains, challenging the linear assumptions invoked in previous studies. The methodology introduced here provides a blueprint for learning other soil processes where high-quality datasets are available but mechanistic understanding is incomplete.

physics.geo-ph↗

arXiv2025

STROKEVISION-BENCH: A Multimodal Video And 2D Pose Benchmark For Tracking Stroke Recovery

Despite advancements in rehabilitation protocols, clinical assessment of upper extremity (UE) function after stroke largely remains subjective, relying heavily on therapist observation and coarse scoring systems. This subjectivity limits the sensitivity of assessments to detect subtle motor improvements, which are critical for personalized rehabilitation planning. Recent progress in computer vision offers promising avenues for enabling objective, quantitative, and scalable assessment of UE motor function. Among standardized tests, the Box and Block Test (BBT) is widely utilized for measuring gross manual dexterity and tracking stroke recovery, providing a structured setting that lends itself well to computational analysis. However, existing datasets targeting stroke rehabilitation primarily focus on daily living activities and often fail to capture clinically structured assessments such as block transfer tasks. Furthermore, many available datasets include a mixture of healthy and stroke-affected individuals, limiting their specificity and clinical utility. To address these critical gaps, we introduce StrokeVision-Bench, the first-ever dedicated dataset of stroke patients performing clinically structured block transfer tasks. StrokeVision-Bench comprises 1,000 annotated videos categorized into four clinically meaningful action classes, with each sample represented in two modalities: raw video frames and 2D skeletal keypoints. We benchmark several state-of-the-art video action recognition and skeleton-based action classification methods to establish performance baselines for this domain and facilitate future research in automated stroke rehabilitation assessment.

eess.IV↗

arXiv2025

AVEX: What Matters for Animal Vocalization Encoding

Bioacoustics, the study of sounds produced by living organisms, plays a vital role in conservation, biodiversity monitoring, and behavioral studies. Many tasks in this field, such as species, individual, and behavior classification and detection, are well-suited to machine learning. However, they often suffer from limited annotated data, highlighting the need for a general-purpose bioacoustic encoder capable of extracting useful representations for diverse downstream tasks. Such encoders have been proposed before, but are often limited in scope due to a focus on a narrow range of species (typically birds), and a reliance on a single model architecture or training paradigm. Moreover, they are usually evaluated on a small set of tasks and datasets. In this work, we present a large-scale empirical study that covers aspects of bioacoustics that are relevant to research but have previously been scarcely considered: training data diversity and scale, model architectures and training recipes, and the breadth of evaluation tasks and datasets. We obtain encoders that are state-of-the-art on the existing and proposed benchmarks. We also identify what matters for training these encoders, such that this work can be extended when more data are available or better architectures are proposed. Specifically, across 26 datasets with tasks including species classification, detection, individual ID, and vocal repertoire discovery, we find self-supervised pre-training followed by supervised post-training on a mixed bioacoustics + general-audio corpus yields the strongest in- and out-of-distribution performance. We show the importance of data diversity in both stages. To support ongoing research and application, we will release the model checkpoints.

cs.SD↗

arXiv2025

gpt-oss-120b & gpt-oss-20b Model Card

We present gpt-oss-120b and gpt-oss-20b, two open-weight reasoning models that push the frontier of accuracy and inference cost. The models use an efficient mixture-of-expert transformer architecture and are trained using large-scale distillation and reinforcement learning. We optimize the models to have strong agentic capabilities (deep research browsing, python tool use, and support for developer-provided functions), all while using a rendered chat format that enables clear instruction following and role delineation. Both models achieve strong results on benchmarks ranging from mathematics, coding, and safety. We release the model weights, inference implementations, tool environments, and tokenizers under an Apache 2.0 license to enable broad use and further research.

cs.CL↗

arXiv2025

Excess risk of heat-related hospitalization associated with temperature and PM2.5 among older adults

Background: With rising temperatures and an aging population, understanding how to prevent heat-related illness among older adults will be increasingly crucial. Despite biological plausibility, no study to date has investigated whether fine particulate matter air pollution (PM2.5) contributes to the risk of hospitalization with a diagnosis code indicating heat-related illness, referred to as heat-related hospitalization. This study aims to fill this gap by investigating the independent and combined effects of temperature and PM2.5 on heat-related hospitalization risk. Methods: We identified Medicare fee-for-service beneficiaries in the contiguous United States who experienced a heat-related hospitalization between 2008 and 2016. Using a case-crossover design and Bayesian conditional logistic regression, we characterized the associations of temperature and PM2.5 with heat-related hospitalization. We then estimated the relative excess risk due to interaction to quantify the additive interaction of simultaneous exposure to heat and PM2.5. Results: We observed 112,969 heat-related hospitalizations. Fixing PM2.5 at the case day median, the odds ratio for increasing temperature from its case day median to the 95th percentile was 1.05 (95% CI: 1.03, 1.06). Fixing temperature at the case day median, the odds ratio for increasing PM2.5 from its median to the 95th percentile was 1.01 (95% CI: 0.99, 1.04). The relative excess risk due to interaction for simultaneous median-to-95th percentile increases in temperature and PM2.5 was 0.03 (95% CI: 0.01, 0.06). Conclusions: Our study is the first to observe synergism between temperature and PM2.5 associated with the risk of heat-related hospitalization. These findings highlight the importance of considering air pollution in effective public health and clinical interventions to prevent heat-related illness.

stat.AP↗

arXiv2025

Cosmic Reionization on Computers: Statistical Properties of the Distributions of Mean Opacities

Quasar absorption lines provide a unique window to the relationship between galaxies and the intergalactic medium during the Epoch of Reionization. In particular, high redshift quasars enable measurements of the neutral hydrogen content of the universe. However, the limited sample size of observed quasar spectra, particularly at the highest redshifts, hampers our ability to fully characterize the intergalactic medium during this epoch from observations alone. In this work, we characterize the distributions of mean opacities of the intergalactic medium in simulations from the Cosmic Reionization on Computers (CROC) project. We find that the distribution of mean opacities along sightlines follows a non-trivial distribution that cannot be easily approximated by a known distribution. When comparing the cumulative distribution function of mean opacities measurements in subsamples of sample sizes similar to observational measurements from the literature, we find consistency between CROC and observations at redshifts $z\lesssim 5.7$. However, at higher redshifts ($z\gtrsim5.7$), the cumulative distribution function of mean opacities from CROC is notably narrower than those from observed quasar sightlines implying that observations probe a systematically more opaque intergalactic medium at higher redshifts than the intergalactic medium in CROC boxes at these same redshifts. This is consistent with previous analyses that indicate that the universe is reionized too early in CROC simulations.

astro-ph.CO↗

arXiv2025

Synthetic data enables context-aware bioacoustic sound event detection

We propose a methodology for training foundation models that enhances their in-context learning capabilities within the domain of bioacoustic signal processing. We use synthetically generated training data, introducing a domain-randomization-based pipeline that constructs diverse acoustic scenes with temporally strong labels. We generate over 8.8 thousand hours of strongly-labeled audio and train a query-by-example, transformer-based model to perform few-shot bioacoustic sound event detection. Our second contribution is a public benchmark of 13 diverse few-shot bioacoustics tasks. Our model outperforms previously published methods, and improves relative to other training-free methods by $64\%$. We demonstrate that this is due to increase in model size and data scale, as well as algorithmic improvements. We make our trained model available via an API, to provide ecologists and ethologists with a training-free tool for bioacoustic sound event detection.

cs.SD↗

arXiv2024

OpenAI o1 System Card

The o1 model series is trained with large-scale reinforcement learning to reason using chain of thought. These advanced reasoning capabilities provide new avenues for improving the safety and robustness of our models. In particular, our models can reason about our safety policies in context when responding to potentially unsafe prompts, through deliberative alignment. This leads to state-of-the-art performance on certain benchmarks for risks such as generating illicit advice, choosing stereotyped responses, and succumbing to known jailbreaks. Training models to incorporate a chain of thought before answering has the potential to unlock substantial benefits, while also increasing potential risks that stem from heightened intelligence. Our results underscore the need for building robust alignment methods, extensively stress-testing their efficacy, and maintaining meticulous risk management protocols. This report outlines the safety work carried out for the OpenAI o1 and OpenAI o1-mini models, including safety evaluations, external red teaming, and Preparedness Framework evaluations.

cs.AI↗

arXiv2024

The effects of different cooling and heating function models on a simulated analog of NGC300

Gas cooling and heating rates are vital components of hydrodynamic simulations. However, they are computationally expensive to evaluate exactly with chemical networks or photoionization codes. We compare two different approximation schemes for gas cooling and heating in an idealized simulation of an isolated galaxy. One approximation is based on a polynomial interpolation of a table of Cloudy calculations, as is commonly done in galaxy formation simulations. The other approximation scheme uses machine learning for the interpolation instead on an analytic function, with improved accuracy. We compare the temperature-density phase diagrams of gas from each simulation run to assess how much the two simulation runs differ. Gas in the simulation using the machine learning approximation is systematically hotter for low-density gas with $-3 \lesssim \log{(n_b/\mathrm{cm}^{-3})} \lesssim -1$. We find a critical curve in the phase diagram where the two simulations have equal amounts of gas. The phase diagrams differ most strongly at temperatures just above and below this critical curve. We compare CII emission rates for collisions with various particles (integrated over the gas distribution function), and find slight differences between the two simulations. Future comparisons with simulations including radiative transfer will be necessary to compare observable quantities like the total CII luminosity.

astro-ph.GA↗

arXiv2024

NatureLM-audio: an Audio-Language Foundation Model for Bioacoustics

Large language models (LLMs) prompted with text and audio have achieved state-of-the-art performance across various auditory tasks, including speech, music, and general audio, showing emergent abilities on unseen tasks. However, their potential has yet to be fully demonstrated in bioacoustics tasks, such as detecting animal vocalizations in large recordings, classifying rare and endangered species, and labeling context and behavior -- tasks that are crucial for conservation, biodiversity monitoring, and animal behavior studies. In this work, we present NatureLM-audio, the first audio-language foundation model specifically designed for bioacoustics. Our training dataset consists of carefully curated text-audio pairs spanning bioacoustics, speech, and music, designed to address the field's limited availability of annotated data. We demonstrate successful transfer of learned representations from music and speech to bioacoustics, and our model shows promising generalization to unseen taxa and tasks. We evaluate NatureLM-audio on a novel benchmark (BEANS-Zero) and it sets a new state of the art on several bioacoustics tasks, including zero-shot classification of unseen species. To advance bioacoustics research, we release our model weights, benchmark data, and open-source the code for training and benchmark data generation and model training.

cs.SD↗

arXiv2024

On the minimum number of radiation field parameters to specify gas cooling and heating functions

Fast and accurate approximations of gas cooling and heating functions are needed for hydrodynamic galaxy simulations. We use machine learning to analyze atomic gas cooling and heating functions computed by Cloudy in the presence of a generalized incident local radiation field. We characterize the radiation field through binned radiation field intensities instead of the photoionization rates used in our previous work. We find a set of 6 energy bins whose intensities exhibit relatively low correlation. We use these bins as features to train machine learning models to predict Cloudy cooling and heating functions at fixed metallicity. We compare the relative SHapley Additive exPlanation (SHAP) value importance of the features. From the SHAP analysis, we identify a feature subset of 3 energy bins ($0.5-1, 1-4$, and $13-16 \, \mathrm{Ry}$) with the largest importance and train additional models on this subset. We compare the mean squared errors and distribution of errors on both the entire training data table and a randomly selected 20% test set withheld from model training. The machine learning models trained with 3 and 6 bins, as well as 3 and 4 photoionization rates, have comparable accuracy everywhere, with errors $\gtrsim 10$ times smaller than for the interpolation table of Gnedin and Hollon (2012). We conclude that 3 energy bins (or 3 analogous photoionization rates: molecular hydrogen photodissociation, neutral hydrogen HI, and fully ionized carbon CVI) are sufficient to characterize the dependence of the gas cooling and heating functions on our assumed incident radiation field model.

astro-ph.GA↗

arXiv2023

Emergence of the Temperature-Density Relation in the Low-Density Intergalactic Medium

We examine the evolution of the phase diagram of the low-density intergalactic medium during the Epoch of Reionization in simulation boxes with varying reionization histories from the Cosmic Reionization on Computers project. The PDF of gas temperature at fixed density exhibits two clear modes: a warm and cold temperature mode, corresponding to the gas inside and outside of ionized bubbles. We find that the transition between the two modes is "universal" in the sense that its timing is accurately parameterized by the value of the volume-weighted neutral fraction for any reionization history. This "universality" is more complex than just a reflection of the fact that ionized gas is warm and neutral gas is cold: it holds for the transition at a fixed value of gas density, and gas at different densities transitions from the cold to the warm mode at different values of the neutral fraction, reflecting a non-trivial relationship between the ionization history and the evolving gas density PDF. Furthermore, the "emergence" of the tight temperature-density relation in the warm mode is also approximately "universally" controlled by the volume-weighted neutral fraction for any reionization history. In particular, the "emergence" of the temperature-density relation (as quantified by the rapid decrease in its width) occurs when the neutral fraction is $10^{-4}\lesssim X_\mathrm{HI} \lesssim10^{-3}$ for any reionization history. Our results indicate that the neutral fraction is a primary quantity controlling the various properties of the temperature-density relation, regardless of reionization history.

astro-ph.CO↗