-
BioDiffusion: A Versatile Diffusion Model for Biomedical Signal Synthesis
Authors:
Xiaomin Li,
Mykhailo Sakevych,
Gentry Atkinson,
Vangelis Metsis
Abstract:
Machine learning tasks involving biomedical signals frequently grapple with issues such as limited data availability, imbalanced datasets, labeling complexities, and the interference of measurement noise. These challenges often hinder the optimal training of machine learning algorithms. Addressing these concerns, we introduce BioDiffusion, a diffusion-based probabilistic model optimized for the sy…
▽ More
Machine learning tasks involving biomedical signals frequently grapple with issues such as limited data availability, imbalanced datasets, labeling complexities, and the interference of measurement noise. These challenges often hinder the optimal training of machine learning algorithms. Addressing these concerns, we introduce BioDiffusion, a diffusion-based probabilistic model optimized for the synthesis of multivariate biomedical signals. BioDiffusion demonstrates excellence in producing high-fidelity, non-stationary, multivariate signals for a range of tasks including unconditional, label-conditional, and signal-conditional generation. Leveraging these synthesized signals offers a notable solution to the aforementioned challenges. Our research encompasses both qualitative and quantitative assessments of the synthesized data quality, underscoring its capacity to bolster accuracy in machine learning tasks tied to biomedical signals. Furthermore, when juxtaposed with current leading time-series generative models, empirical evidence suggests that BioDiffusion outperforms them in biomedical signal generation quality.
△ Less
Submitted 27 January, 2024; v1 submitted 12 January, 2024;
originally announced January 2024.
-
The Effects of Signal-to-Noise Ratio on Generative Adversarial Networks Applied to Marine Bioacoustic Data
Authors:
Georgia Atkinson,
Nick Wright,
A. Stephen McGough,
Per Berggren
Abstract:
In recent years generative adversarial networks (GANs) have been used to supplement datasets within the field of marine bioacoustics. This is driven by factors such as the cost to collect data, data sparsity and aid preprocessing. One notable challenge with marine bioacoustic data is the low signal-to-noise ratio (SNR) posing difficulty when applying deep learning techniques such as GANs. This wor…
▽ More
In recent years generative adversarial networks (GANs) have been used to supplement datasets within the field of marine bioacoustics. This is driven by factors such as the cost to collect data, data sparsity and aid preprocessing. One notable challenge with marine bioacoustic data is the low signal-to-noise ratio (SNR) posing difficulty when applying deep learning techniques such as GANs. This work investigates the effect SNR has on the audio-based GAN performance and examines three different evaluation methodologies for GAN performance, yielding interesting results on the effects of SNR on GANs, specifically WaveGAN.
△ Less
Submitted 22 December, 2023;
originally announced December 2023.
-
SANA: Cross-Species Prediction of Gene Ontology GO Annotations via Topological Network Alignment
Authors:
Siyue Wang,
Giles R. S. Atkinson,
Wayne B. Hayes
Abstract:
Topological network alignment aims to align two networks node-wise in order to maximize the observed common connection (edge) topology between them. The topological alignment of two Protein-Protein Interaction (PPI) networks should thus expose protein pairs with similar interaction partners allowing, for example, the prediction of common Gene Ontology (GO) terms. Unfortunately, no network alignmen…
▽ More
Topological network alignment aims to align two networks node-wise in order to maximize the observed common connection (edge) topology between them. The topological alignment of two Protein-Protein Interaction (PPI) networks should thus expose protein pairs with similar interaction partners allowing, for example, the prediction of common Gene Ontology (GO) terms. Unfortunately, no network alignment algorithm based on topology alone has been able to achieve this aim, though those that include sequence similarity have seen some success. We argue that this failure of topology alone is due to the sparsity and incompleteness of the PPI network data of almost all species, which provides the network topology with a small signal-to-noise ratio that is effectively swamped when sequence information is added to the mix. Here we show that the weak signal can be detected using multiple stochastic samples of "good" topological network alignments, which allows us to observe regions of the two networks that are robustly aligned across multiple samples. The resulting Network Alignment Frequency (NAF) strongly correlates with GO-based Resnik semantic similarity and enables the first successful cross-species predictions of GO terms based on topology-only network alignments. Our best predictions have an AUPR of about 0.4, which is competitive with state-of-the-art algorithms, even when there is no observable sequence similarity and no known homology relationship. While our results provide only a "proof of concept" on existing network data, we hypothesize that predicting GO terms from topology-only network alignments will become increasingly practical as the volume and quality of PPI network data increase.
△ Less
Submitted 26 April, 2022;
originally announced April 2022.
-
Structural features of microvascular networks trigger blood-flow oscillations
Authors:
Yaron Ben-Ami,
George W. Atkinson,
Joe M. Pitt-Francis,
Philip K. Maini,
Helen M. Byrne
Abstract:
We analyse mathematical models in order to understand how microstructural features of vascular networks may affect blood-flow dynamics, and to identify particular characteristics that promote the onset of self-sustained oscillations. By focusing on a simple three-node motif, we predict that network "redundancy", in the form of a redundant vessel connecting two main flow-branches, together with dif…
▽ More
We analyse mathematical models in order to understand how microstructural features of vascular networks may affect blood-flow dynamics, and to identify particular characteristics that promote the onset of self-sustained oscillations. By focusing on a simple three-node motif, we predict that network "redundancy", in the form of a redundant vessel connecting two main flow-branches, together with differences in haemodynamic resistance in the branches, can promote the emergence of oscillatory dynamics. We use existing mathematical descriptions for blood rheology and haematocrit splitting at vessel branch-points to construct our flow model; we combine numerical simulations and stability analysis to study the dynamics of the three-node network and its relation to the system's multiple steady-state solutions. While, for the case of equal inlet-pressure conditions, a "trivial" equilibrium solution with no flow in the redundant vessel always exists, we find that it is not stable when other, stable, steady-state attractors exist. In turn, these "nontrivial" steady-state solutions may undergo a Hopf bifurcation into an oscillatory state. We use the branch diameter ratio, together with the inlet haematocrit rate, to construct a two-parameter stability diagram that delineates regimes in which such oscillatory dynamics exist. We show that flow oscillations in this network geometry are only possible when the branch diameters are sufficiently different to allow for a sufficiently large flow in the redundant vessel, which acts as the driving force of the oscillations. These microstructural properties, which were found to promote oscillatory dynamics, could be used to explore sources of flow instability in biological microvascular networks.
△ Less
Submitted 22 June, 2022; v1 submitted 26 February, 2022;
originally announced March 2022.
-
Measure Twice, Cut Once: Quantifying Bias and Fairness in Deep Neural Networks
Authors:
Cody Blakeney,
Gentry Atkinson,
Nathaniel Huish,
Yan Yan,
Vangelis Metris,
Ziliang Zong
Abstract:
Algorithmic bias is of increasing concern, both to the research community, and society at large. Bias in AI is more abstract and unintuitive than traditional forms of discrimination and can be more difficult to detect and mitigate. A clear gap exists in the current literature on evaluating the relative bias in the performance of multi-class classifiers. In this work, we propose two simple yet effe…
▽ More
Algorithmic bias is of increasing concern, both to the research community, and society at large. Bias in AI is more abstract and unintuitive than traditional forms of discrimination and can be more difficult to detect and mitigate. A clear gap exists in the current literature on evaluating the relative bias in the performance of multi-class classifiers. In this work, we propose two simple yet effective metrics, Combined Error Variance (CEV) and Symmetric Distance Error (SDE), to quantitatively evaluate the class-wise bias of two models in comparison to one another. By evaluating the performance of these new metrics and by demonstrating their practical application, we show that they can be used to measure fairness as well as bias. These demonstrations show that our metrics can address specific needs for measuring bias in multi-class classification.
△ Less
Submitted 8 October, 2021;
originally announced October 2021.
-
Quantum Enhanced Precision Estimation of Transmission with Bright Squeezed Light
Authors:
G. S. Atkinson,
E. J. Allen,
G. Ferranti,
A. R. McMillan,
J. C. F. Matthews
Abstract:
Squeezed light enables measurements with sensitivity beyond the quantum noise limit (QNL) for optical techniques such as spectroscopy, gravitational wave detection, magnetometry and imaging. Precision of a measurement -- as quantified by the variance of repeated estimates -- has also been enhanced beyond the QNL using squeezed light. However, sub-QNL sensitivity is not sufficient to achieve sub-QN…
▽ More
Squeezed light enables measurements with sensitivity beyond the quantum noise limit (QNL) for optical techniques such as spectroscopy, gravitational wave detection, magnetometry and imaging. Precision of a measurement -- as quantified by the variance of repeated estimates -- has also been enhanced beyond the QNL using squeezed light. However, sub-QNL sensitivity is not sufficient to achieve sub-QNL precision. Furthermore, demonstrations of sub-QNL precision in estimating transmission have been limited to picowatts of probe power. Here we demonstrate simultaneous enhancement of precision and sensitivity to beyond the QNL for estimating modulated transmission with a squeezed amplitude probe of 0.2 mW average (25 W peak) power, which is 8 orders of magnitude above the power limitations of previous sub-QNL precision measurements of transmission. Our approach enables measurements that compete with the optical powers of current classical techniques, but have both improved precision and sensitivity beyond the classical limit.
△ Less
Submitted 24 February, 2021; v1 submitted 25 September, 2020;
originally announced September 2020.
-
NDD20: A large-scale few-shot dolphin dataset for coarse and fine-grained categorisation
Authors:
Cameron Trotter,
Georgia Atkinson,
Matt Sharpe,
Kirsten Richardson,
A. Stephen McGough,
Nick Wright,
Ben Burville,
Per Berggren
Abstract:
We introduce the Northumberland Dolphin Dataset 2020 (NDD20), a challenging image dataset annotated for both coarse and fine-grained instance segmentation and categorisation. This dataset, the first release of the NDD, was created in response to the rapid expansion of computer vision into conservation research and the production of field-deployable systems suited to extreme environmental condition…
▽ More
We introduce the Northumberland Dolphin Dataset 2020 (NDD20), a challenging image dataset annotated for both coarse and fine-grained instance segmentation and categorisation. This dataset, the first release of the NDD, was created in response to the rapid expansion of computer vision into conservation research and the production of field-deployable systems suited to extreme environmental conditions -- an area with few open source datasets. NDD20 contains a large collection of above and below water images of two different dolphin species for traditional coarse and fine-grained segmentation. All data contained in NDD20 was obtained via manual collection in the North Sea around the Northumberland coastline, UK. We present experimentation using standard deep learning network architecture trained using NDD20 and report baselines results.
△ Less
Submitted 27 May, 2020;
originally announced May 2020.
-
The Northumberland Dolphin Dataset: A Multimedia Individual Cetacean Dataset for Fine-Grained Categorisation
Authors:
Cameron Trotter,
Georgia Atkinson,
Matthew Sharpe,
A. Stephen McGough,
Nick Wright,
Per Berggren
Abstract:
Methods for cetacean research include photo-identification (photo-id) and passive acoustic monitoring (PAM) which generate thousands of images per expedition that are currently hand categorised by researchers into the individual dolphins sighted. With the vast amount of data obtained it is crucially important to develop a system that is able to categorise this quickly. The Northumberland Dolphin D…
▽ More
Methods for cetacean research include photo-identification (photo-id) and passive acoustic monitoring (PAM) which generate thousands of images per expedition that are currently hand categorised by researchers into the individual dolphins sighted. With the vast amount of data obtained it is crucially important to develop a system that is able to categorise this quickly. The Northumberland Dolphin Dataset (NDD) is an on-going novel dataset project made up of above and below water images of, and spectrograms of whistles from, white-beaked dolphins. These are produced by photo-id and PAM data collection methods applied off the coast of Northumberland, UK. This dataset will aid in building cetacean identification models, reducing the number of human-hours required to categorise images. Example use cases and areas identified for speed up are examined.
△ Less
Submitted 7 August, 2019;
originally announced August 2019.
-
Efficient deterministic giant photon phase shift from a single charged quantum dot
Authors:
P. Androvitsaneas,
A. B. Young,
J. M. Lennon,
C. Schneider,
S. Maier,
J. J. Hinchliff,
G. Atkinson,
M. Kamp,
S. Höfling,
J. G. Rarity,
R. Oulton
Abstract:
Solid-state quantum emitters have long been recognised as the ideal platform to realize integrated quantum photonic technologies. We use a self-assembled negatively charged QD in a low Q-factor photonic micropillar to demonstrate for the first time a key figure of merit for deterministic switching and spin-photon entanglement: a shift in phase of an input single photon of $>90^{o}$ with values of…
▽ More
Solid-state quantum emitters have long been recognised as the ideal platform to realize integrated quantum photonic technologies. We use a self-assembled negatively charged QD in a low Q-factor photonic micropillar to demonstrate for the first time a key figure of merit for deterministic switching and spin-photon entanglement: a shift in phase of an input single photon of $>90^{o}$ with values of up to $2π/3$ ($120^{o}$) demonstrated. This $>π/2$ ($90^{o}$) measured value represents an important threshold: above this value input photons interact with the emitter deterministically. A deterministic photon-emitter interaction is the only viable scalable means to achieve several vital functionalities not possible in linear optics such as quantum switches and entanglement gates. Our experimentally determined value is limited by mode mismatch between the input laser and the cavity, QD spectral fluctuations and spin relaxation. We determine that up to $80\%$ of the collected photons have interacted with the QD and undergone a phase shift of $π$.
△ Less
Submitted 10 May, 2017; v1 submitted 9 September, 2016;
originally announced September 2016.
-
Experimental demonstration of acoustic wave induced magnetization switching of dipole coupled magnetostrictive nanomagnets for ultralow power computing
Authors:
Vimal Sampath,
Noel DSouza,
Gary Atkinson,
Supriyo Bandyopadhyay,
Jayasimha Atulasimha
Abstract:
We report nanomagnetic switching with Acoustic Waves (AW) launched from interdigitated electrodes that modulate the stress anisotropy of elliptical cobalt nanoscale magnetostrictive magnets (340 nm x 270 nm x 12 nm) delineated on 128 degree Y-cut lithium niobate. The dipole-coupled nanomagnet pairs are in a single-domain state and are initially magnetized along the major axis of the ellipse, with…
▽ More
We report nanomagnetic switching with Acoustic Waves (AW) launched from interdigitated electrodes that modulate the stress anisotropy of elliptical cobalt nanoscale magnetostrictive magnets (340 nm x 270 nm x 12 nm) delineated on 128 degree Y-cut lithium niobate. The dipole-coupled nanomagnet pairs are in a single-domain state and are initially magnetized along the major axis of the ellipse, with their magnetizations parallel to each other. The magnetizations of nanomagnets having lower shape anisotropy are reversed upon acoustic wave propagation. Thereafter, the magnetization of these nanomagnets remains in the reversed state and demonstrate non-volatility. This executes a 'NOT' operation. This proof of acoustic wave induced magnetic state reversal in dipole-coupled nanomagnets implementing a 'NOT' gate operation could potentially lead to the development of extremely energy-efficient nanomagnetic logic. Furthermore, fabrication complexity is reduced immensely due to the absence of individual contacts to the nanomagnets, leading to lower energy dissipation
△ Less
Submitted 3 June, 2016;
originally announced June 2016.
-
Switching the magnetization of magnetostrictive nanomagnets from single-domain to non-volatile vortex states with a surface acoustic wave
Authors:
Vimal Sampath,
Noel ďSouza,
Dhritiman Bhattacharya,
Gary M. Atkinson,
Supriyo Bandyopadhyay,
Jayasimha Atulasimha
Abstract:
We report manipulation of the magnetic states of elliptical cobalt magnetostrictive nanomagnets (of nominal dimensions ~ 340 nm x 270 nm x 12 nm) delineated on bulk 128° Y-cut lithium niobate with Surface Acoustic Waves (SAWs) launched from interdigitated electrodes. Isolated nanomagnets that are initially magnetized to a single domain state with magnetization pointing along the major axis of the…
▽ More
We report manipulation of the magnetic states of elliptical cobalt magnetostrictive nanomagnets (of nominal dimensions ~ 340 nm x 270 nm x 12 nm) delineated on bulk 128° Y-cut lithium niobate with Surface Acoustic Waves (SAWs) launched from interdigitated electrodes. Isolated nanomagnets that are initially magnetized to a single domain state with magnetization pointing along the major axis of the ellipse are driven into a vortex state by surface acoustic waves that modulate the stress anisotropy of these nanomagnets. The nanomagnets remain in the vortex state until they are reset by a strong magnetic field to the initial single domain state, making the vortex state non-volatile. This phenomenon is modeled and explained using a micromagnetic framework and could lead to the development of extremely energy efficient magnetization switching methodologies.
△ Less
Submitted 9 January, 2016;
originally announced January 2016.
-
Reconnection in the Centre of the Near-Earth Plasma Sheet: Control by Plasma Drifts and Magnetic Flux Convection close to the Earth
Authors:
Gerald Atkinson
Abstract:
Convection inside the magnetosphere can be regarded as the transfer of magnetic flux and plasma from the magnetotail into the nightside and then convection and drift from there to the dayside. As in many fluid flow situations, the downstream flow (the convection and drift to the dayside) is an important boundary condition for the upstream flow (reconnection and inflow from the tail). Tail-like f…
▽ More
Convection inside the magnetosphere can be regarded as the transfer of magnetic flux and plasma from the magnetotail into the nightside and then convection and drift from there to the dayside. As in many fluid flow situations, the downstream flow (the convection and drift to the dayside) is an important boundary condition for the upstream flow (reconnection and inflow from the tail). Tail-like flux tubes in the near-earth plasma sheet block inflow from the tail between reconnection events. However, the westward drift of the energetic ions (curvature and gradient drift and ExB drift) removes plasma energy from near midnight causing deflation and earthward flow in the equatorial plane. (The process can be described as the westward drift of the eastern of the plasma sheet / partial ring current.) The earthward flow removes magnetic flux and plasma from the center of the thick plasma sheet and creates a thin current sheet, which, in turn, favors the onset of reconnection. The X line forms just tailward of the earthward-flow region. Initially, reconnection occurs by the earthward propagation of the X line and the growth of a magnetic island on its tailward side, since there is no downtail flow. Subsequently, the system evolves towards lobe merging, and the X line propagates tailward consistent with the standard picture of tail reconnection. The east-west width of the local-time slot in which inflow and reconnection occur is determined by the scale-length of the decrease in plasma energy at the eastern end of the drifting ions. It is expected that the reconnection slot follows the westward drift of the eastern end of the partial ring current. The model appears to be consistent with bursty bulk flows, the westward traveling surge and many other observed properties of substorms.
△ Less
Submitted 17 December, 2004; v1 submitted 6 November, 2004;
originally announced November 2004.
