Search

We propose a general pipeline to automate the extraction of labels from radiology reports using large language models, which we validate on spinal MRI reports. The efficacy of our labelling method is measured on five distinct conditions: spinal cancer, stenosis, spondylolisthesis, cauda equina compression and herniation. Using open-source models, our method equals or surpasses GPT-4 on a held-out set of reports. Furthermore, we show that the extracted labels can be used to train imaging models to classify the identified conditions in the accompanying MR scans. All classifiers trained using automated labels achieve comparable performance to models trained using scans manually annotated by clinicians. Code can be found at https://github.com/robinyjpark/AutoLabelClassifier., Comment: Accepted to Medical Image Computing and Computer Assisted Intervention (MICCAI 2024, Spotlight). 11 pages plus appendix

This paper describes an efficient, open source PyTorch implementation of the Spectral Transform Unit. We investigate sequence prediction tasks over several modalities including language, robotics, and simulated dynamical systems. We find that for the same parameter count, the STU and its variants outperform the Transformer as well as other leading state space models across various modalities.

PLATO (PLAnetary Transits and Oscillations of stars) is ESA's M3 mission designed to detect and characterise extrasolar planets and perform asteroseismic monitoring of a large number of stars. PLATO will detect small planets (down to <2 R_(Earth)) around bright stars (<11 mag), including terrestrial planets in the habitable zone of solar-like stars. With the complement of radial velocity observations from the ground, planets will be characterised for their radius, mass, and age with high accuracy (5 %, 10 %, 10 % for an Earth-Sun combination respectively). PLATO will provide us with a large-scale catalogue of well-characterised small planets up to intermediate orbital periods, relevant for a meaningful comparison to planet formation theories and to better understand planet evolution. It will make possible comparative exoplanetology to place our Solar System planets in a broader context. In parallel, PLATO will study (host) stars using asteroseismology, allowing us to determine the stellar properties with high accuracy, substantially enhancing our knowledge of stellar structure and evolution. The payload instrument consists of 26 cameras with 12cm aperture each. For at least four years, the mission will perform high-precision photometric measurements. Here we review the science objectives, present PLATO's target samples and fields, provide an overview of expected core science performance as well as a description of the instrument and the mission profile at the beginning of the serial production of the flight cameras. PLATO is scheduled for a launch date end 2026. This overview therefore provides a summary of the mission to the community in preparation of the upcoming operational phases.

We study monolayer WSe2 using ultrafast electron diffraction. We introduce an approach to quantitatively extract atomic-site-specific information, providing an element-specific view of incoherent atomic vibrations following femtosecond excitation. Via differences between W and Se vibrations, we identify stages in the nonthermal evolution of the lattice. Combined with a calculated phonon dispersion, this element specificity enables us to identify a long-lasting overpopulation of specific optical phonons, and to interpret the stages as energy transfer processes between specific phonon groups. These results demonstrate the appeal of resolving element-specific vibrational information in the ultrafast time domain., Comment: 13 pages, 3 figures

Optical manipulation of magnetism holds promise for future ultrafast spintronics, especially with lanthanides and their huge, localized 4f magnetic moments. These moments interact indirectly via the conduction electrons (RKKY exchange), influenced by interatomic orbital overlap, and the conduction electron susceptibility. Here, we study this influence in a series of 4f antiferromagnets, GdT2Si2 (T=Co, Rh, Ir), using ultrafast resonant X-ray diffraction. We observe a twofold increase in ultrafast angular momentum transfer between the materials, originating from modifications in the conduction electron susceptibility, as confirmed by first-principles calculations.

Understanding the limits of rocky planet habitability is one of the key goals of current and future exoplanet characterization efforts. An intrinsic concept of rocky planet habitability is the Habitable Zone. To date, the most widely used estimates of the Habitable Zone are based on cloud-free, one-dimensional (vertical) radiative-convective climate model calculations. However, recent three-dimensional global climate modeling efforts have revealed that rocky planet habitability is strongly impacted by radiative cloud feedbacks, where computational expense and model limitations can prevent these tools from exploring the limits of habitability across the full range of parameter space. We leverage a patchy cloud one-dimensional radiative-convective climate model with parameterized cloud microphysics to investigate Inner Edge limits to the Habitable Zone for main sequence stars ($T_{\rm eff}$ = 2600 -7200K). We find that Inner Edge limits to the Habitable Zone can be 3.3 and 4.7 times closer than previous cloud-free estimates for Earth- and super-Earth-sized worlds, respectively, depending on bulk cloud parameters (e.g., fractional cloudiness and sedimentation efficiency). These warm, moist Inner Edge climates are expected to have extensive cloud decks that could mute deep atmosphere spectral features. To aid in rocky planet characterization studies, we identify the potential of using $\rm{CO_{\rm 2}}$ absorption features in transmission spectroscopy as a means of quantifying cloud deck height and cloud sedimentation efficiency. Moist greenhouse climates may represent key yet poorly understood states of habitable planets for which continued study will uncover new insights into the search and characterization of habitable worlds., Comment: 33 pages, 12 figures, submitted, community comments welcome

Child Impact Statements (CIS) are instrumental in helping to foreground the concerns and needs of minor community members who are too young to vote and often unable to advocate for themselves politically. While many politicians and policymakers assert they make decisions in the best interests of children, they often lack the necessary information to meaningfully accomplish this. CISs are akin to Environmental Impact Statements in that both give voice to constituents who are often under-represented in policymaking. This paper highlights an interdisciplinary collaboration between Social Science and Computer Science to create a CIS tool for policymakers and community members in Shelby County, TN. Furthermore, this type of collaboration is fruitful beyond the scope of the CIS tool. Social scientists and computer scientists can leverage their complementary skill sets in data management and data interpretation for the benefit of their communities, advance scientific knowledge, and bridge disciplinary divides within the academy.

Chemical disequilibrium quantified via available free energy has previously been proposed as a potential biosignature. However, exoplanet biosignature remote sensing work has not yet investigated how observational uncertainties impact the ability to infer a life-generated available free energy. We pair an atmospheric retrieval tool to a thermodynamics model to assess the detectability of chemical disequilibrium signatures of Earth-like exoplanets, emphasizing the Proterozoic Eon where atmospheric abundances of oxygen-methane disequilibrium pairs may have been relatively high. Retrieval model studies applied across a range of gas abundances revealed that order-of-magnitude constraints on disequilibrium energy are achieved with simulated reflected-light observations at the high abundance scenario and signal-to-noise ratios (50) while weak constraints are found at moderate SNRs (20\,--\,30) for med\,--\,low abundance cases. Furthermore, the disequilibrium energy constraints are improved by modest thermal information encoded in water vapor opacities at optical and near-infrared wavelengths. These results highlight how remotely detecting chemical disequilibrium biosignatures can be a useful and metabolism-agnostic approach to biosignature detection., Comment: Nature Astronomy. Supplementary information see https://zenodo.org/records/10093798 For Source Data see https://zenodo.org/records/8335447

Soft robotics is an emerging technology in which engineers create flexible devices for use in a variety of applications. In order to advance the wide adoption of soft robots, ensuring their trustworthiness is essential; if soft robots are not trusted, they will not be used to their full potential. In order to demonstrate trustworthiness, a specification needs to be formulated to define what is trustworthy. However, even for soft robotic grippers, which is one of the most mature areas in soft robotics, the soft robotics community has so far given very little attention to formulating specifications. In this work, we discuss the importance of developing specifications during development of soft robotic systems, and present an extensive example specification for a soft gripper for pick-and-place tasks for grocery items. The proposed specification covers both functional and non-functional requirements, such as reliability, safety, adaptability, predictability, ethics, and regulations. We also highlight the need to promote verifiability as a first-class objective in the design of a soft gripper., Comment: Updated the Standards subsection of paper. 9 pages, 2 figures, 1 table, 34 references

This work introduces a simple deep-learning based method to delineate contours by `walking' along learnt unit vector fields. We demonstrate the effectiveness of our pipeline on the unique case of open contours on the task of delineating the sacroiliac joints (SIJs) in spinal MRIs. We show that: (i) 95% of the time the average root mean square error of the predicted contour against the original ground truth is below 4.5 pixels (2.5mm for a standard T1-weighted SIJ MRI), and (ii) the proposed method is better than the baseline of regressing vertices or landmarks of contours., Comment: IEEE International Symposium on Biomedical Imaging (ISBI) 2023

This paper explores training medical vision-language models (VLMs) -- where the visual and language inputs are embedded into a common space -- with a particular focus on scenarios where training data is limited, as is often the case in clinical datasets. We explore several candidate methods to improve low-data performance, including: (i) adapting generic pre-trained models to novel image and text domains (i.e. medical imaging and reports) via unimodal self-supervision; (ii) using local (e.g. GLoRIA) & global (e.g. InfoNCE) contrastive loss functions as well as a combination of the two; (iii) extra supervision during VLM training, via: (a) image- and text-only self-supervision, and (b) creating additional positive image-text pairs for training through augmentation and nearest-neighbour search. Using text-to-image retrieval as a benchmark, we evaluate the performance of these methods with variable sized training datasets of paired chest X-rays and radiological reports. Combined, they significantly improve retrieval compared to fine-tuning CLIP, roughly equivalent to training with the data. A similar pattern is found in the downstream task classification of CXR-related conditions with our method outperforming CLIP and also BioVIL, a strong CXR VLM benchmark, in the zero-shot and linear probing settings. We conclude with a set of recommendations for researchers aiming to train vision-language models on other medical imaging modalities when training data is scarce. To facilitate further research, we will make our code and models publicly available., Comment: Accepted to MIDL 2023

We present Thistle, a fully functional vector database. Thistle is an entry into the domain of latent knowledge use in answering search queries, an ongoing research topic at both start-ups and search engine companies. We implement Thistle with several well-known algorithms, and benchmark results on the MS MARCO dataset. Results help clarify the latent knowledge domain as well as the growing Rust ML ecosystem.

The Reinforcement Learning field is strong on achievements and weak on reapplication; a computer playing GO at a super-human level is still terrible at Tic-Tac-Toe. This paper asks whether the method of training networks improves their generalization. Specifically we explore core quality diversity algorithms, compare against two recent algorithms, and propose a new algorithm to deal with shortcomings in existing methods. Although results of these methods are well below the performance hoped for, our work raises important points about the choice of behavior criterion in quality diversity, the interaction of differential and evolutionary training methods, and the role of offline reinforcement learning and randomized learning in evolutionary search.

Sentence simplification tends to focus on the generic simplification of sentences by making them more readable and easier to understand. This paper provides a dataset aimed at training models that perform subject aware sentence simplifications rather than simplifying sentences as a whole. We also test models on that dataset which are inspired by model architecture used in abstractive summarization. We hand generated portions of the data and augment the dataset by further manipulating those hand written simplifications. Our results show that data-augmentation, data-masking, and model architecture choices used in summarization provide a solid baseline for comparison on subject aware simplification.

The behaviours of a swarm are not explicitly engineered. Instead, they are an emergent consequence of the interactions of individual agents with each other and their environment. This emergent functionality poses a challenge to safety assurance. The main contribution of this paper is a process for the safety assurance of emergent behaviour in autonomous robotic swarms called AERoS, following the guidance on the Assurance of Machine Learning for use in Autonomous Systems (AMLAS). We explore our proposed process using a case study centred on a robot swarm operating a public cloakroom., Comment: 12 pages, 11 figures

Clouds are ubiquitous\, -- \,they arise for every solar system planet that possesses an atmosphere and have also been suggested as a leading mechanism for obscuring spectral features in exoplanet observations. As exoplanet observations continue to improve, there is a need for efficient and general planetary climate models that appropriately handle the possible cloudy atmospheric environments that arise on these worlds. We generate a new 1D radiative-convective terrestrial planet climate model that self-consistently handles patchy clouds through a parameterized microphysical treatment of condensation and sedimentation processes. Our model is general enough to recreate Earth's atmospheric radiative environment without over-parameterization, while also maintaining a simple implementation that is applicable to a wide range of atmospheric compositions and physical planetary properties. We first validate this new 1D patchy cloud radiative-convective climate model by comparing it to Earth thermal structure data and to existing climate and radiative transfer tools. We produce partially-clouded Earth-like climates with cloud structures that are representative of deep tropospheric convection and are adequate 1D representations of clouds within rocky planet atmospheres. After validation against Earth, we then use our partially clouded climate model and explore the potential climates of super-Earth exoplanets with secondary nitrogen-dominated atmospheres which we assume are abiotic. We also couple the partially clouded climate model to a full-physics, line-by-line radiative transfer model and generate high-resolution spectra of simulated climates. These self-consistent climate-to-spectral models bridge the gap between climate modeling efforts and observational studies of rocky worlds., Comment: 28 pages, 15 figures, submitted; community comments welcome

Hexagonal YMnO$_{3}$ is a prototype antiferromagnet which exhibits multiferroic behavior with the ferroelectric and magnetic transitions occurring at different temperatures. We observe an out-of-plane canting of the Mn$^{3+}$ magnetic moments using resonant X-ray diffraction (RXD) in a single crystal of this material. These canted moments result in the symmetry-forbidden (0,0,1) magnetic Bragg reflection, which is observed at the Mn $L_{2,3}$ absorption edges. We also observe an unexpected difference in the RXD spectral shapes at different temperatures. Using ab initio calculations, we explore the possibility that this behavior arises due to the interference between scattering from the canted magnetic moments and parity-odd atomic multipoles on the Mn$^{3+}$ ions.

The microscopic origin of slow carrier cooling in lead-halide perovskites remains debated, and has direct implications for applications. Slow carrier cooling has been attributed to either polaron formation or a hot-phonon bottleneck effect at high excited carrier densities (> 10$^{18}$ cm$^{-3}$). These effects cannot be unambiguously disentangled from optical experiments alone. However, they can be distinguished by direct observations of ultrafast lattice dynamics, as these effects are expected to create qualitatively distinct fingerprints. To this end, we employ femtosecond electron diffraction and directly measure the sub-picosecond lattice dynamics of weakly confined CsPbBr$_3$ nanocrystals following above-gap photo-excitation. The data reveal a light-induced structural distortion appearing on a time scale varying between 380 fs to 1200 fs depending on the excitation fluence. We attribute these dynamics to the effect of exciton-polarons on the lattice, and the slower dynamics at high fluences to slower hot carrier cooling, which slows down the establishment of the exciton-polaron population. Further analysis and simulations show that the distortion is consistent with motions of the [PbBr$_3$]$^{-}$ octahedral ionic cage, and closest agreement with the data is obtained for Pb-Br bond lengthening. Our work demonstrates how direct studies of lattice dynamics on the sub-picosecond timescale can discriminate between competing scenarios, thereby shedding light on the origin of slow carrier cooling in lead-halide perovskites.

Periodic tilings can store information if individual tiles are decorated to lower their symmetry. Truchet tilings - the broad family of space-filling arrangements of such tiles - offer an efficient mechanism of visual data storage related to that used in barcodes and QR codes. Here, we show that the crystalline metal-organic framework [OZn$_4$][1,3-benzenedicarboxylate]$_3$ (TRUMOF-1) is an atomic-scale realisation of a complex three-dimensional Truchet tiling. Its crystal structure consists of a periodically-arranged assembly of identical zinc-containing clusters connected uniformly in a well-defined but disordered fashion to give a topologically aperiodic microporous network. We suggest that this unusual structure emerges as a consequence of geometric frustration in the chemical building units from which it is assembled.

The ultrafast manipulation of magnetic order due to optical excitation is governed by the intricate flow of energy and momentum between the electron, lattice and spin subsystems. While various models are commonly employed to describe these dynamics, a prominent example being the microscopic three temperature model (M3TM), systematic, quantitative comparisons to both the dynamics of energy flow and magnetic order are scarce. Here, we apply a M3TM to the ultrafast magnetic order dynamics of the layered antiferromagnet GdRh$_2$Si$_2$. The femtosecond dynamics of electronic temperature, surface ferromagnetic order, and bulk antiferromagnetic order were explored at various pump fluences employing time- and angle-resolved photoemission spectroscopy and time-resolved resonant magnetic soft x-ray diffraction, respectively. After optical excitation, both the surface ferromagnetic order and the bulk antiferromagnetic order dynamics exhibit two-step demagnetization behaviors with two similar timescales (<1 ps, ~10 ps), indicating a strong exchange coupling between localized 4f and itinerant conduction electrons. Despite a good qualitative agreement, the M3TM predicts larger demagnetization than our experimental observation, which can be phenomenologically described by a transient, fluence-dependent increased N\'eel temperature. Our results indicate that effects beyond a mean-field description have to be considered for a quantitative description of ultrafast magnetic order dynamics., Comment: 19 pages, 10 figures

This paper proposes a novel transformer-based model architecture for medical imaging problems involving analysis of vertebrae. It considers two applications of such models in MR images: (a) detection of spinal metastases and the related conditions of vertebral fractures and metastatic cord compression, (b) radiological grading of common degenerative changes in intervertebral discs. Our contributions are as follows: (i) We propose a Spinal Context Transformer (SCT), a deep-learning architecture suited for the analysis of repeated anatomical structures in medical imaging such as vertebral bodies (VBs). Unlike previous related methods, SCT considers all VBs as viewed in all available image modalities together, making predictions for each based on context from the rest of the spinal column and all available imaging modalities. (ii) We apply the architecture to a novel and important task: detecting spinal metastases and the related conditions of cord compression and vertebral fractures/collapse from multi-series spinal MR scans. This is done using annotations extracted from free-text radiological reports as opposed to bespoke annotation. However, the resulting model shows strong agreement with vertebral-level bespoke radiologist annotations on the test set. (iii) We also apply SCT to an existing problem: radiological grading of inter-vertebral discs (IVDs) in lumbar MR scans for common degenerative changes.We show that by considering the context of vertebral bodies in the image, SCT improves the accuracy for several gradings compared to previously published model., Comment: Pre-print of paper accepted to MICCAI 2022. 15 pages, 7 figures

As autonomous systems (AS) increasingly become part of our daily lives, ensuring their trustworthiness is crucial. In order to demonstrate the trustworthiness of an AS, we first need to specify what is required for an AS to be considered trustworthy. This roadmap paper identifies key challenges for specifying for trustworthiness in AS, as identified during the "Specifying for Trustworthiness" workshop held as part of the UK Research and Innovation (UKRI) Trustworthy Autonomous Systems (TAS) programme. We look across a range of AS domains with consideration of the resilience, trust, functionality, verifiability, security, and governance and regulation of AS and identify some of the key specification challenges in these domains. We then highlight the intellectual challenges that are involved with specifying for trustworthiness in AS that cut across domains and are exacerbated by the inherent uncertainty involved with the environments in which AS need to operate., Comment: Accepted version of paper. 13 pages, 1 table, 1 figure