1,817 research outputs found
The construction of Karen Karnak: The multi-author-function
No full textThis thesis is situated within the comparatively recent developments of Web 2.0 and the emergence of interactive WikiMedia, and explores the mode of authorship within a Read/Write culture compared to that of a Read/Only tradition. The hypothesis of this study is that the role of the audience has become merged with the author, and as such, represents new functions and attributes, distinct from a more conventional concept of authorship, in which the roles of audience and author are more separate. Read/Write and participatory culture, as defined by this study, is focused on collaboration, and includes the influences of D.I.Y. culture, Open-Source practices and the production of text by multiple authors. Multi-authorship presents a re-thinking of several concepts which support the notion of the individual author, since the focus of multi-authorship is not on attribution and ownership of a finished text, but on the continued malleability of a text. Modes of multi-authorship, demonstrated in the use of the pseudonyms Alan Smithee and Karen Eliot, represent declarative authors whose names signify multiple origins, whilst concurrently indicating a distinct body of work. The function of these names form an important context to this study, since primary research involves the construction of an experimental mode of multi-authorship utilising WikiMedia technology and the interaction of thirty nine participants, who are invited to create a body of work under the collective pseudonym Karen Karnak. The data generated by this experiment is analysed using aspects of Michel Foucault's author-function to identify and determine power structures inherent in the WikiMedia context. The interplay of power structures, including concepts such as identity, ownership and the body of work, affect the resulting mode of authorship and contribute to the construction of Karen Karnak, suggesting further areas of research into the emerging multi-author
FSD-FS
No full textFSD-FS is a publicly-available database of human labelled sound events for few-shot learning. It spans across 143 classes obtained from the AudioSet Ontology and contains 43805 raw audio files collected from the FSD50K. FSD-FS is curated at the Centre for Digital Music, Queen Mary University of London.
Citation
If you use the FSD-FS dataset, please cite our paper and FSD50K.
@article{liang2022learning,
title={Learning from Taxonomy: Multi-label Few-Shot Classification for Everyday Sound Recognition},
author={Liang, Jinhua and Phan, Huy and Benetos, Emmanouil},
journal={arXiv preprint arXiv:2212.08952},
year={2022}
}
@ARTICLE{9645159, author={Fonseca, Eduardo and Favory, Xavier and Pons, Jordi and Font, Frederic and Serra, Xavier}, journal={IEEE/ACM Transactions on Audio, Speech, and Language Processing}, title={FSD50K: An Open Dataset of Human-Labeled Sound Events}, year={2022}, volume={30}, number={}, pages={829-852}, doi={10.1109/TASLP.2021.3133208}}
About FSD-FS
FSD-FS is an open database for multi-label few-shot audio classification containing 143 classes drawn from the FSD50K. It also inherits the AudioSet Ontology. FSD-FS follows the ratio 7:2:1 to split classes into base, validation, and evaluation sets, so there are 98 classes in the base set, 30 classes in the validation set, and 15 classes in the evaluation set (More details can be found in our paper).
LICENSE
FSD-FS are released in Creative Commons (CC) licenses. Same as FSD50K, each clip has its own license as defined by the clip uploader in Freesound, some of them requiring attribution to their original authors and some forbidding further commercial reuse. For more details, ones can refer to the link.
FILES
FSD-FS are organised in the structure:
root
|
└─── dev_base
|
└─── dev_val
|
└─── eval
REFERENCES AND LINKS
[1] Gemmeke, Jort F., et al. "Audio set: An ontology and human-labeled dataset for audio events." 2017 IEEE international conference on acoustics, speech and signal processing (ICASSP). IEEE, 2017. [paper] [link]
[2] Fonseca, Eduardo, et al. "Fsd50k: an open dataset of human-labeled sound events." IEEE/ACM Transactions on Audio, Speech, and Language Processing 30 (2021): 829-852. [paper] [code
Artificial neural networks fighting real neural decline: a systematic review of AI in Alzheimer’s research
No full textData availability:
All data used in this systematic review were obtained from publicly available sources and peer-reviewed journal articles. The selection process is outlined in the PRISMA flow diagram included in the manuscript. A complete list of included studies is available in the Zotero library used for this review. This library can be shared upon reasonable request to the corresponding author. Any new data generated during this study can be obtained upon request.Alzheimer’s disease (AD) is a major global health challenge, with Artificial Intelligence (AI) increasingly recognized as a transformative tool for early detection, disease progression modeling, and therapeutic discovery. This systematic review, conducted in accordance with PRISMA guidelines, analyzed 156 peer-reviewed studies published between 2010 and 2024, identified from four major databases (Scopus, PubMed, Web of Science, IEEE Xplore). A particular emphasis was placed on multimodal approaches that integrate neuroimaging, genetics, biomarkers, and clinical data to improve accuracy and translational value. To organize this fragmented field, we introduce a novel Layered Framework that categorizes AI applications into four domains: Early Detection, Disease Progression Modeling, Therapeutic Discovery, and Real-World Integration. In addition, we applied ARIMA-based forecasting to project research trajectories through 2030, which revealed generative models and transformer architectures as the fastest-growing and most promising methodologies. The review highlights substantial advances in early detection and multimodal fusion, particularly through deep learning, while also identifying persistent challenges such as limited model generalizability, ethical concerns, and underexplored clinical implementation. Addressing these barriers will require multi-cohort validation, interpretable AI, and equity-driven model development. By consolidating evidence and forecasting future directions, this review provides a roadmap for accelerating precision-driven innovations in Alzheimer’s care.This work was supported by the Engineering and Physical Sciences Research Council
Tissue architecture of the anterior pituitary showing the epithelial cell cords with hormonal cells and folliculo-stellate (FS) cells, the capillaries (C) with fenestrated endothelial cells (EC) and connective tissue (CT)
No full text<p><b>Copyright information:</b></p><p>Taken from "Paracrinicity: The Story of 30 Years of Cellular Pituitary Crosstalk"</p><p></p><p>Journal of Neuroendocrinology 2008;20(1):1-70.</p><p>Published online Jan 2008</p><p>PMCID:PMC2229370.</p><p>© 2008 The Author. Journal Compilation © 2008 Blackwell Publishing Ltd</p> The cell cords are a cluster of endocrine cells surrounding an aggregate of FS cells that make a follicle (F). FS cells also make a meshwork between the hormonal cells, making junctions among each other (thick lines) and extending foot processes (f) ending on the basal membrane (BM) in the periphery of the cord. The cords are surrounded by BM, which may have extensions between some cells. A second BM surrounds the capillary vessels and between these two some connective tissue resides. Small and larger lacunae are present between hormonal cells. Paracrine substances may circulate from cell-to-cell but also could be released in these lacunae and reach more remote places. FS cells make gap junctions mostly among each other, but occasionally also with some hormonal cells. Hormonal cells can make interdigitations with FS cells (small arrows) to favour cell-to-cell communication. Adapted from Vila-Porcile ()
Adaptive pulse compression for transform-limited 15-fs high-energy pulse generation
No full textWe demonstrate the use of a deformable-mirror pulse shaper, combined with an evolutionary optimization algorithm, to correct high-order residual phase aberrations in a 1-mJ, 1-kHz, 15-fs laser amplifier. Frequency resolved optical gating measurements reveal that the output pulse duration of 15.2 fs is within our measurement error of the theoretical transform limit. This technique significantly reduces the pulse duration and the temporal prepulse energy of the pulse while increasing the peak intensity by 26%. It is demonstrated, for what is believed to be the first time, that the problem of pedestals in laser amplifiers can be addressed by spectral-domain correction
Self-compression of 4.9 µm pulses to sub-40 fs with 2 mJ energy in Zinc Sulfide
No full textNonlinear self-compression of few-cycle multi-mJ pulses at 4.9 µm in ZnS is presented. 80 fs input pulses are compressed to 37 fs with 2.1 mJ energy at a 1 kHz repetition rate. © 2024 The Author(s
Sub-100 fs mode-locked erbium-doped fiber laser using a 45°-tilted fiber grating
No full textWe demonstrate generation of sub-100 fs pulses at 1.5 μm in a mode-locked erbium-doped fiber laser using a 45°-tilted fiber grating element. The laser features a genuine all-fiber configuration. Based on the unique polarization properties of the 45°-tilted fiber grating, we managed to produce sub-100 fs laser pulses through proper dispersion management. To the best of our knowledge, this is the shortest pulse generated from modelocked lasers with fiber gratings. The output pulse has an average power of 8 mW, with a repetition rate of 47.8 MHz and pulse energy of 1.68 nJ. The performance of laser also matches well the theoretical simulations. © 2013 Optical Society of America
Sub-40 fs Er:fiber laser
No full textSimulation and experiment results are presented for sub-40 fs pulse generation in Er: fiber ring lasers. (C) 2010 Optical Society of Americahttp://gateway.webofknowledge.com/gateway/Gateway.cgi?GWVersion=2&SrcApp=PARTNER_APP&SrcAuth=LinksAMR&KeyUT=WOS:000295612401079&DestLinkType=FullRecord&DestApp=ALL_WOS&UsrCustomerID=8e1609b174ce4e31116a60747a720701Engineering, Electrical & ElectronicOpticsPhysics, AppliedCPCI-S(ISTP)
Time-domain investigation of the ro-vibrational CH<sub>4</sub> v<sub>2</sub> Raman spectrum via fs/ps CRS for combustion diagnostics
No full textWith the onset of climate change it is crucial to eliminate emissions, particularly for aviation reliant on gas turbine combustion. Computational models aid combustor improvement, but experimental data needed for validation is difficult to obtain. Fortunately, laser diagnostics can now achieve non-intrusive, high-resolution combustion measurements, especially with hybrid femtosecond/picosecond (fs/ps) coherent Raman scattering (CRS). In this thesis, the goal was to develop hybrid fs/ps CRS spectroscopy for the CH4 ν2 vibrational mode. Methane merits investigation as a climate forcer alongside potential use in carbon-neutral hydrogen production. Time-resolved CRS was performed for collision-independent, collisional dephasing, and laminar CH4-air flame measurements. A time-domain model well-reproduced experiments, along with extracting molecule-specific ν2 Q-branch dephasing linewidth parameters. Additionally, fundamental physio-chemical processes were observed in a spectral molecular fingerprint region across the CH4-air flame: CH4 dissociation, H2 synthesis and consumption, mass diffusion, and fuel/oxidizer mixing.Aerospace Engineerin
- …
