Edinburgh DataShare
Not a member yet
7718 research outputs found
Sort by
Multi-view Disentanglement for Reinforcement Learning with Multiple Cameras
No full textThe performance of image-based Reinforcement Learning (RL) agents can vary depending on the position of the camera used to capture the images. Training on multiple cameras simultaneously, including a first-person egocentric camera, can leverage information from different camera perspectives to improve the performance of RL. However, hardware constraints may limit the availability of multiple cameras in real-world deployment. Additionally, cameras may become damaged in the real-world preventing access to all cameras that were used during training. To overcome these hardware constraints, we propose Multi-View Disentanglement (MVD), which uses multiple cameras to learn a policy that is robust to a reduction in the number of cameras to generalise to any single camera from the training set. Our approach is a self-supervised auxiliary task for RL that learns a disentangled representation from multiple cameras, with a shared representation that is aligned across all cameras to allow generalisation to a single camera, and a private representation that is camera-specific. We show experimentally that an RL agent trained on a single third-person camera is unable to learn an optimal policy in many control tasks; but, our approach, benefiting from multiple cameras during training, is able to solve the task using only the same single third-person camera.The performance of image-based Reinforcement Learning (RL) agents can vary depending on the position of the camera used to capture the images. Training on multiple cameras simultaneously, including a first-person egocentric camera, can leverage information from different camera perspectives to improve the performance of RL. However, hardware constraints may limit the availability of multiple cameras in real-world deployment. Additionally, cameras may become damaged in the real-world preventing access to all cameras that were used during training. To overcome these hardware constraints, we propose Multi-View Disentanglement (MVD), which uses multiple cameras to learn a policy that is robust to a reduction in the number of cameras to generalise to any single camera from the training set. Our approach is a self-supervised auxiliary task for RL that learns a disentangled representation from multiple cameras, with a shared representation that is aligned across all cameras to allow generalisation to a single camera, and a private representation that is camera-specific. We show experimentally that an RL agent trained on a single third-person camera is unable to learn an optimal policy in many control tasks; but, our approach, benefiting from multiple cameras during training, is able to solve the task using only the same single third-person camera.
This is the data for the experiment results in the paper 'Multi-view Disentanglement for Reinforcement Learning for Multiple Cameras' (https://arxiv.org/abs/2404.14064). These files contain the evaluation returns for all MVD seeds used to create Figures 4 and 5 in the paper. Further details are provided in the README file
SAL - Salawati (ISO 369-3: xmx, Glottocode: made1253) segmented files
No full textThis dataset contains segmented audio files from several (not all) of the raw recordings in the Salawati (ISO 369-3: xmx, Glottocode: made1253) collection. The full collection is available here: https://datashare.ed.ac.uk/handle/10283/8672. These segmented files are designed to facilitate searches for a particular word or construction in the Salawati collection. Full information on how to navigate these segmented files using the filenames is given in the Documentation file. The audio files in this dataset come from recordings that were made during two trips: the first were recorded between 29 November and 17 December 2019, in Kalobo and Yenenas villages (Salawati and Batanta islands, Raja Ampat regency, Southwest Papua province, Indonesia); and the second between 14–16 March 2023, in Solol village (Salawati island, Raja Ampat regency, Southwest Papua province, Indonesia). This dataset relates to the British Academy Postdoctoral Fellowship 'Synchronic and diachronic investigations in Raja Ampat-South Halmahera, a little-known subbranch of Austronesian' (PF19\100004); additional fieldwork funds were provided by a British Academy Small Grant (SG1920\100342) and a Moray Endowment Fund from the University of Edinburgh. Further deposits are available in the related DataShare sub-community (https://datashare.ed.ac.uk/handle/10283/8573)
SUPERSEDED: Primary data for manuscript on a homologue of lymphostatin in Chlamydia pecorum
No full text## ## This item has been replaced by the one which can be found at https://hdl.handle.net/10283/9002 ## Pathogens frequently produce proteins to evade or inhibit host immune responses. One such protein is lymphostatin from attaching and effacing Escherichia coli (also known as lymphocyte inhibitory factor A; LifA), which influences intestinal colonization and inhibits mitogen- and antigen-activated proliferation of T lymphocytes and pro-inflammatory cytokine synthesis. Here, we report the cloning, purification and characterization of a LifA homologue from Chlamydia pecorum. The predicted 382 KDa protein (CPE2_0552) exhibited 36 % identity and 55 % similarity over 3171 amino acids to lymphostatin from enteropathogenic E. coli strain E2348/69. CPE2_0552 shares glycosyltransferase and cysteine protease motifs required for lymphostatin activity, including similarity in the tertiary structure of these domains predicted by AlphaFold 3. Purified CPE2_0552 exhibited a surface envelope similar to that of lymphostatin when analyzed by electron microscopy. CPE2_0552 inhibited concanavalin A-stimulated proliferation of bovine T cells in a concentration-dependent manner, with an inhibitory dose 50 (ID50) of 990 pg/mL. This was 70-fold higher than the ID50 of E. coli E2348/69 lymphostatin tested in parallel on T cells from the same donors (14 ± 4 pg/mL), but was similar to another LifA homologue from E. coli O157:H7 (ToxB). Moreover, CPE2_0552 inhibited the secretion of interferon gamma (IFN?), a key cytokine that influences the outcome of Chlamydia infections. At the concentrations at which CPE2_0552 inhibited T lymphocyte proliferation and IFN? secretion, negligible cytotoxicity was observed after 72 h of stimulation. Our study indicates that E. coli lymphostatin belongs to a wider family of lymphocyte-inhibitory molecules that exist in distantly related bacterial pathogens
Raw red counts from a CRISPR screen for host factors required for influenza virus replication
No full textRaw data for the CRISPR screen we conducted in collaboration with Bo Li, John Doench and Nir Hacohen can be downloaded here. Please refer to the manuscript for full details: [doi.org/10.1038/s41467-019-13965-x]
Host dependency factors that are required for influenza A virus infection may serve as therapeutic targets as the virus is less likely to bypass them under drug-mediated selection pressure. Previous attempts to identify host factors have produced largely divergent results, with few overlapping hits across different studies. Here, we perform a genome-wide CRISPR/Cas9 screen and devise a new approach, meta-analysis by information content (MAIC) to systematically combine our results with prior evidence for influenza host factors. MAIC out-performs other meta-analysis methods when using our CRISPR screen as validation data. We validate the host factors, WDR7, CCDC115 and TMEM199, demonstrating that these genes are essential for viral entry and regulation of V-type ATPase assembly. We also find that CMTR1, a human mRNA cap methyltransferase, is required for efficient viral cap snatching and regulation of a cell autonomous immune response, and provides synergistic protection with the influenza endonuclease inhibitor Xofluza
CFD cases and data for 'Unsteady Load Mitigation through Passive Pitch' published on Journal of Fluids & Structures
No full textMitigation of load fluctuations due to flow unsteadiness is critical in a broad range of applications, including wind/tidal turbines, and aerial/underwater vehicles. While the use of active control systems is an established practice in engineering, passive systems are not well understood, and the limits of their efficacy are yet to be ascertained. To this end, the present study aims to provide new insights into the effectiveness of passive pitching in the mitigation of lift fluctuations in the most demanding case of fast, high-amplitude variations of the free stream speed and direction. We perform fluid-structure interaction simulations of a two-dimensional free-to-pitch rigid foil.
Our study reveals that the lift amplitude of the force fluctuations can be decreased by at least two-thirds through passive pitching. The efficacy of the unsteady load mitigation is only weakly dependent on the exact pitching axis location, and the optimal position is upstream and close to the axis of the foil. These results may inform the design of passive control systems of wind/tidal turbines and aerial/underwater vehicles and provide new insights into interpreting the control strategy of natural flyers such as insects and birds
Experimental and simulation study of reverse micelles formed by Aerosol-OT and water in non-polar solvents
No full textText files contain raw molecular dynamics (MD) simulation and small-angle neutron scattering (SANS) data presented in the article. For Figures 1 and 2, MD and SANS in the file name indicate the origin of the data. For all other figures, the data are from MD simulations only. All other data in the article are either given in a table in the article, or generated using equations and parameters given in the article.See README.tx
Unraveling the Regimes of Interfacial Thermal Conductance at a Solid/Liquid Interface
No full textThe interfacial thermal conductance at a solid/liquid interface (G) exhibits an exponential-to-linear cross-over with increasing solid/liquid interaction strength, previously attributed to the relative strength of solid/liquid to liquid/liquid interactions. Instead, using a simple Lennard-Jones setup, our molecular simulations reveal that this cross-over occurs due to the onset of solidification in the interfacial liquid at high solid/liquid interaction strengths. This solidification subsequently influences interfacial energy transport, leading to the cross-over in G. We use the overlap between the spectrally decomposed heat fluxes of the interfacial solid and liquid to pinpoint when “solid-like energy transport” within the interfacial liquid emerges. We also propose a novel decomposition of G into: i) the conductance right at the solid/liquid interface and ii) the conductance of the nanoscale interfacial liquid region. We demonstrate that the rise of solid-like energy transport within the interfacial liquid influences the relative magnitude of these conductances, which in turn dictates when the cross-over occurs. Our results can aid engineers in optimizing G at realistic interfaces, critical to designing effective cooling solutions for electronics among other applications. This data set contains the MD simulation files and post-processing codes to reproduce our results. The dataset is related to the publication by Abdullah El-Rifai, Sree Hari Perumanath Dharmapalan, Matthew K. Borg and Rohit Pillai (2024), "Unraveling the Regimes of Interfacial Thermal Conductance at a Solid/Liquid Interface", Journal of Physical Chemistry C, https://doi.org/10.1021/acs.jpcc.4c00536File structure described in README.tx
Land surface phenology and biodiversity in Zambian woodlands
No full textSeasonal foliage display (leaf phenology) is a key determinant of ecosystem function. Variation in land surface phenology, observed via space-borne remote sensing, can be explained at broad spatial scales by climate, but we lack understanding of how vegetation structure and floristic diversity mediates these relationships. This lack of understanding hampers our ability to predict changes in phenology and therefore ecosystem function, in light of rapid ongoing shifts in biodiversity and ecosystem structure due to land use and climate change.
We combined a network of 619 vegetation monitoring sites across deciduous woodlands in Zambia with land surface phenology metrics to investigate the role of tree species diversity, composition and vegetation structure on patterns of land surface phenology, including the phenomenon of pre-rain green-up.
Tree species diversity was associated with earlier pre-rain green-up, a longer growing season, and greater cumulative foliage production. Independent of diversity, proportional abundance of Detarioideae species (subfamily of Fabaceae) was associated with a longer growing season by facilitating earlier pre-rain green-up. Woodland stands with larger trees green up earlier, suggesting access to deep groundwater reserves. Senescence metrics showed variation among sites, but were not well explained by precipitation, temperature, structure or diversity.
Synthesis: Tree diversity, composition and structure are co-determinants of seasonal patterns of foliage display in tropical deciduous woodlands, as measured via land surface phenology, at regional scale. Our study identifies both a niche complementarity effect whereby diverse woodlands exhibit longer growing seasons, as well as a selection effect whereby detarioid species drive earlier pre-rain green-up, providing insights into the mechanisms underlying the biodiversity ecosystem function relationship in tropical deciduous woodlands. We stress the importance of considering biotic controls on ecosystem functioning in the next generation of earth-system models predicting the response of communities to global change
BAT - Batta (Glottocode: bata1295) segmented files
No full textThis dataset contains segmented audio files from several (not all) of the raw recordings in the Batta (Glottocode: bata1295) collection. The full collection is available here: https://datashare.ed.ac.uk/handle/10283/8614. These segmented files are designed to facilitate searches for a particular word or construction in the Batta collection. Full information on how to navigate these segmented files using the filenames is given in the Documentation file. The audio files in this dataset come from recordings that were made during two trips: the first were recorded between 9 and 19 December 2019, the second between 9–11 March 2023, all in Yenenas village (Batanta island, Raja Ampat regency, Southwest Papua province, Indonesia). This dataset relates to the British Academy Postdoctoral Fellowship 'Synchronic and diachronic investigations in Raja Ampat-South Halmahera, a little-known subbranch of Austronesian' (PF19\100004); additional fieldwork funds were provided by a British Academy Small Grant (SG1920\100342) and a Moray Endowment Fund from the University of Edinburgh. Further deposits are available in the related DataShare sub-community (https://datashare.ed.ac.uk/handle/10283/8573)
Tracking the Conical Intersection Dynamics for the Photoinduced Jahn-Teller Switch of a Mn(III) Complex
No full textRaw data sets for transient absorption measurements and calculations for the article "Tracking the Conical Intersection Dynamics for the Photoinduced Jahn-Teller Switch of a Mn(III) Complex" accepted for publication in Chemical Science