BackgroundSeizure prediction can increase independence and allow preventative treatment for patients with epilepsy. We present a proof-of-concept for a seizure prediction system that is accurate, fully automated, patient-specific, and tunable to an individual's needs.MethodsIntracranial electroencephalography (iEEG) data of ten patients obtained from a seizure advisory system were analyzed as part of a pseudoprospective seizure prediction study. First, a deep learning classifier was trained to distinguish between preictal and interictal signals. Second, classifier performance was tested on held-out iEEG data from all patients and benchmarked against the performance of a random predictor. Third, the prediction system was tuned so sensitivity or time in warning could be prioritized by the patient. Finally, a demonstration of the feasibility of deployment of the prediction system onto an ultra-low power neuromorphic chip for autonomous operation on a wearable device is provided.ResultsThe prediction system achieved mean sensitivity of 69% and mean time in warning of 27%, significantly surpassing an equivalent random predictor for all patients by 42%.ConclusionThis study demonstrates that deep learning in combination with neuromorphic hardware can provide the basis for a wearable, real-time, always-on, patient-specific seizure warning system with low power consumption and reliable long-term performance.
Deep learning technology is uniquely suited to analyse neurophysiological signals such as the electroencephalogram (EEG) and local field potentials (LFP) and promises to outperform traditional machine-learning based classification and feature extraction algorithms. Furthermore, novel cognitive computing platforms such as IBM's recently introduced neuromorphic TrueNorth chip allow for deploying deep learning techniques in an ultra-low power environment with a minimum device footprint. Merging deep learning and TrueNorth technologies for real-time analysis of brain-activity data at the point of sensing will create the next generation of wearables at the intersection of neurobionics and artificial intelligence.
Zinc oxide nanocrystals were prepared in ethanol and spin-cast to form semiconductor nanocrystal thin films that were thermally annealed at temperatures between 100 and 800 °C. Particle size, monodispersity, and film porosity were determined by X-ray diffraction, ultraviolet-visible absorption spectroscopy, and spectroscopic ellipsometry, respectively. Film porosity rapidly decreased above 400 °C, from 32% to 26%, which coincided with a change in electronic properties. Above 400 °C, the ZnO electron mobility, determined from FET transfer characteristics, increased from 10-3 to 10-1 cm2 V s-1, while the surface resistivity, determined from electrical impedance, decreased from 107 to 103 Ω m over the same temperature range. Below the densification point, nanoparticle core resistivity was found to increase from 104 to 106 Ω m, which is caused by the increasing polydispersity in the quantized energy levels of the nanocrystals. From 100 to 800 °C, crystallite size was found to increase from 5 to 18 nm in diameter. The surface resistance was decreased dramatically by passivation with butane thiol
Surface related and intrinsic exciton recombination dynamics in ZnO nanoparticles synthesized by a sol-gel method Appl. Phys. Lett. 102, 013109 (2013); 10.1063/1.4774002 Effects of Mg incorporation on the optical properties of ZnO prepared by the sol-gel method Temperature-dependent photoluminescence of nanocrystalline ZnO thin films grown on Si (100) substrates by the sol-gel process Appl. Phys. Lett. 86, 131910 (2005); 10.1063/1.1891288Stable photoluminescence of zinc oxide quantum dots in silica nanoparticles matrix prepared by the combined sol-gel and spray drying method Semiconductor quantum dots (QDs) are used to dope wide-bandgap chalogenide glasses via sol-gel processing. Such chalcogenides enhance surface passivation of the quantum dots, as evidenced by the increased PL emissions of both core and core shell species used, while a ZnO glass leads to irreversible oxidation of the embedded quantum dots. The embedded QDs are photostable.
wileyonlinelibrary.com COMMUNICATIONamplitude of the PBG. Compared with the other approaches used to tune the dielectric contrast of 3D structures, [14][15][16][17][18][19][20][21] our AC-LBL method offers signifi cant advantages because the use of charged compounds ensures that the coating penetrates deep and uniformly inside 3D complex structures. Moreover, this low cost, fast and room-temperature method can be applied to different coating materials, such as quantum dots and other nanocrystals. [22][23][24] It is also possible to functionalize the particles used to coat the PhCs in order to add new functionalities to the structures.The 3D PhCs considered in this work are 1-srs networks, intricate structures inspired by the gyroid minimal surface discovered by Schoen in 1970, [ 25 ] and observed in butterfl y wing scales ( Figure 2 a). [ 26,27 ] The space group of the 1-srs network is I 4 1 32 and it can be infl ated with constant pressure according to the Young-Laplace equation to form the single gyroid structure of constant mean curvature. [ 28,29 ] Due to their unique geometrical properties, gyroids host a rich variety of optical phenomena, from linear and circular dichroism, [ 29 ] to optical activity, [ 30 ] and the recent demonstration of Weyl points. [ 8 ] PhCs with gyroid symmetry have placed the most signifi cant demands on high refractive index. [ 8 ] On the polymer templates, we deposited high refractive-index PbS thin fi lms with thickness controllable at a nanometre-scale, in order to increase the effective refractive index of the structure. The protocol is optimized for the controlled deposition of PbS nanocubes over the entire surface of a 3D PhC fabricated with a zirconium-based organic-inorganic photosensitive material. [ 31 ] Combining simulation and experimental study, we confi rm that the effective refractive index of the PhCs is increased by ≈40% after only six deposition cycles. This leads to an increase in the width and the strength of the PBG by more than 90% and 40%, respectively.The AC-LBL method consists of the alternating deposition of two oppositely charged species: the negatively charged PbS nanocubes and a positively charged polymer. This approach has the advantage of creating a tunable and homogenous coating on the entire surface of the 3D PhC, overcoming the issue related to directional coating techniques, such as thermal evaporation or chemical vapour deposition. To create the micron-scale corecladding structure illustrated in Figure 1 , we fi rst fabricated a polymer template (Figure 2 b,c) using a galvo-dithered DLW method, [ 6 ] which can create 3D PhCs with cubic symmetry, good mechanical strength, and high resolution ( Figure S1, Supporting Information).To boost the effective refractive index of the polymeric 1-srs PhCs, we adopt PbS nanocubes (Figure 2 e,f) for the coating because this material can be rapidly synthesised with high yield, high refractive index, and good transparency in the visible and near-infrared (NIR) wavelength region. [ 32,33 ] The PbS
Deep Neural Networks (DNN) achieve human level performance in many image analytics tasks but DNNs are mostly deployed to GPU platforms that consume a considerable amount of power. New hardware platforms using lower precision arithmetic achieve drastic reductions in power consumption. More recently, brain-inspired spiking neuromorphic chips have achieved even lower power consumption, on the order of milliwatts, while still offering real-time processing. However, for deploying DNNs to energy efficient neuromorphic chips the incompatibility between continuous neurons and synaptic weights of traditional DNNs, discrete spiking neurons and synapses of neuromorphic chips need to be overcome. Previous work has achieved this by training a network to learn continuous probabilities, before it is deployed to a neuromorphic architecture, such as IBM TrueNorth Neurosynaptic System, by random sampling these probabilities. The main contribution of this paper is a new learning algorithm that learns a TrueNorth configuration ready for deployment. We achieve this by training directly a binary hardware crossbar that accommodates the TrueNorth axon configuration constrains and we propose a different neuron model. Results of our approach trained on electroencephalogram (EEG) data show a significant improvement with previous work (76% vs 86% accuracy) while maintaining state of the art performance on the MNIST handwritten data set.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.