3D-integrated multilayered physical reservoir array for learning and forecasting time-series information.

A wide reservoir computing system is an advanced architecture composed of multiple reservoir layers in parallel, which enables more complex and diverse internal dynamics for multiple time-series information processing. However, its hardware implementation has not yet been realized due to the lack of a high-performance physical reservoir and the complexity of fabricating multiple stacks. Here, we achieve a proof-of-principle demonstration of such hardware made of a multilayered three-dimensional stacked 3 × 10 × 10 tungsten oxide memristive crossbar array, with which we further realize a wide physical reservoir computing for efficient learning and forecasting of multiple time-series data. Because a three-layer structure allows the seamless and effective extraction of intricate three-dimensional local features produced by various temporal inputs, it can readily outperform two-dimensional based approaches extensively studied previously. Our demonstration paves the way for wide physical reservoir computing systems capable of efficiently processing multiple dynamic time-series information.

Filament-free memristors for computing.

Memristors have attracted increasing attention due to their tremendous potential to accelerate data-centric computing systems. The dynamic reconfiguration of memristive devices in response to external electrical stimuli can provide highly desirable novel functionalities for computing applications when compared with conventional complementary-metal-oxide-semiconductor (CMOS)-based devices. Those most intensively studied and extensively reviewed memristors in the literature so far have been filamentary type memristors, which typically exhibit a relatively large variability from device to device and from switching cycle to cycle. On the other hand, filament-free switching memristors have shown a better uniformity and attractive dynamical properties, which can enable a variety of new computing paradigms but have rarely been reviewed. In this article, a wide range of filament-free switching memristors and their corresponding computing applications are reviewed. Various junction structures, switching properties, and switching principles of filament-free memristors are surveyed and discussed. Furthermore, we introduce recent advances in different computing schemes and their demonstrations based on non-filamentary memristors. This Review aims to present valuable insights and guidelines regarding the key computational primitives and implementations enabled by these filament-free switching memristors.

Glutamyl-prolyl-tRNA synthetase 1 coordinates early endosomal anti-inflammatory AKT signaling.

The AKT signaling pathway plays critical roles in the resolution of inflammation. However, the underlying mechanisms of anti-inflammatory regulation and signal coordination remain unclear. Here, we report that anti-inflammatory AKT signaling is coordinated by glutamyl-prolyl-tRNA synthetase 1 (EPRS1). Upon inflammatory activation, AKT specifically phosphorylates Ser999 of EPRS1 in the cytoplasmic multi-tRNA synthetase complex, inducing release of EPRS1. EPRS1 compartmentalizes AKT to early endosomes via selective binding to the endosomal membrane lipid phosphatidylinositol 3-phosphate and assembles an AKT signaling complex specific for anti-inflammatory activity. These events promote AKT activation-mediated GSK3ß phosphorylation, which increase anti-inflammatory cytokine production. EPRS1-deficient macrophages do not assemble the early endosomal complex and consequently exacerbate inflammation, decreasing the survival of EPRS1-deficient mice undergoing septic shock and ulcerative colitis. Collectively, our findings show that the housekeeping protein EPRS1 acts as a mediator of inflammatory homeostasis by coordinating compartment-specific AKT signaling.

A Learning-Rate Modulable and Reliable TiOx Memristor Array for Robust, Fast, and Accurate Neuromorphic Computing.

Realization of memristor-based neuromorphic hardware system is important to achieve energy efficient bigdata processing and artificial intelligence in integrated device system-level. In this sense, uniform and reliable titanium oxide (TiOx ) memristor array devices are fabricated to be utilized as constituent device element in hardware neural network, representing passive matrix array structure enabling vector-matrix multiplication process between multisignal and trained synaptic weight. In particular, in situ convolutional neural network hardware system is designed and implemented using a multiple 25 × 25 TiOx memristor arrays and the memristor device parameters are developed to bring global constant voltage programming scheme for entire cells in crossbar array without any voltage tuning peripheral circuit such as transistor. Moreover, the learning rate modulation during in situ hardware training process is successfully achieved due to superior TiOx memristor performance such as threshold uniformity (≈2.7%), device yield (> 99%), repetitive stability (≈3000 spikes), low asymmetry value of ≈1.43, ambient stability (6 months), and nonlinear pulse response. The learning rate modulable fast-converging in situ training based on direct memristor operation shows five times less training iterations and reduces training energy compared to the conventional hardware in situ training at ≈95.2% of classification accuracy.

Mechanical and Biocompatibility Properties of Sintered Titanium Powder for Mimetic 3D-Printed Bone Scaffolds.

A composite comprising Ti and NaCl powders was sintered similar to a three-dimensional (3D)-printed patient-customized artificial bone scaffold. Additionally, a proper microstructure of the mimetic scaffold and the optimum processing parameters for its development were analyzed. The mechanical properties of the metal-based porous-structured framework used as an artificial bone scaffold were an optimum replacement for the human bone. Thus, it was confirmed that patient-customized scaffolds could be manufactured via 3D printing. The 3D-printed mimetic specimens were fabricated by a powder-sintering method using Ti for the metal parts, NaCl as the pore former, and polylactic acid as the biodegradable binder. Scanning electron microscopy (SEM) images showed that pores were formed homogeneously, while X-ray computed tomography confirmed that open pores were generated. The porosity and pore size distribution were measured using a mercury porosimeter, while the flexural strength and flexural elastic modulus were calculated using the three-point bending test. Based on these measurements, a pore-former content of 15 vol % optimized the density and flexural strength to 2.52 g cm-2 and 283 MPa, respectively, similar to those of the actual iliac bone. According to the 3D-printing production method, a selective laser-sintering process was applied for the fabrication of the mimetic specimen, and it was determined that the microstructure and properties similar to those of previous metal specimens could be achieved in the as-prepared specimen. Additionally, a decellularized extracellular matrix (dECM) was used to coat the surfaces and interiors of the specimens for evaluating their biocompatibilities. SEM image analysis indicated that the adipose-derived stem cells grew evenly inside the pores of the coated specimens, as compared with the bulky Ti specimens without the dECM coating. The doubling time at 65% was measured at 72, 75, and 83 h for specimens with pore-former contents of 5, 10, and 15 vol %, respectively. The doubling time without the pore former was 116 h. As compared with the specimens without the pore former (73 h), 15% of the dECM-coated specimens showed a doubling time of 64%, measured at 47 h.

Vibrio vulnificus PlpA facilitates necrotic host cell death induced by the pore forming MARTX toxin.

Opportunistic pathogen Vibrio vulnificus causes severe systemic infection in humans with high mortality. Although multiple exotoxins have been characterized in V. vulnificus, their interactions and potential synergistic roles in pathogen-induced host cell death have not been investigated previously. By employing a series of multiple exotoxin deletion mutants, we investigated whether specific exotoxins of the pathogen functioned together to achieve severe and rapid necrotic cell death. Human epithelial cells treated with V. vulnificus with a plpA deletion background exhibited an unusually prolonged cell blebbing, suggesting the importance of PlpA, a phospholipase A2, in rapid necrotic cell death by this pathogen. Additional deletion of the rtxA gene encoding the multifunctional autoprocessing repeats-in-toxin (MARTX) toxin did not result in necrotic cell blebs. However, if the rtxA gene was engineered to produce an effector-free MARTX toxin, the cell blebbing was observed, indicating that the pore forming activity of the MARTX toxin is sufficient, but the MARTX toxin effector domains are not necessary, for the blebbing. When a recombinant PlpA was treated on the blebbed cells, the blebs were completely disrupted. Consistent with this, MARTX toxin-pendent rapid release of cytosolic lactate dehydrogenase was significantly delayed in the plpA deletion background. Mutations in other exotoxins such as elastase, cytolysin/hemolysin, and/or extracellular metalloprotease did not affect the bleb formation or disruption. Together, these findings indicate that the pore forming MARTX toxin and the phospholipase A2, PlpA, cooperate sequentially to achieve rapid necrotic cell death by inducing cell blebbing and disrupting the blebs, respectively.

Flexible Neural Network Realized by the Probabilistic SiOx Memristive Synaptic Array for Energy-Efficient Image Learning.

The human brain's neural networks are sparsely connected via tunable and probabilistic synapses, which may be essential for performing energy-efficient cognitive and intellectual functions. In this sense, the implementation of a flexible neural network with probabilistic synapses is a first step toward realizing the ultimate energy-efficient computing framework. Here, inspired by the efficient threshold-tunable and probabilistic rod-to-rod bipolar synapses in the human visual system, a 16 × 16 crossbar array comprising the vertical form of gate-tunable probabilistic SiOx memristive synaptic barristor utilizing the Si/graphene heterojunction is designed and fabricated. Controllable stochastic switching dynamics in this array are achieved via various input voltage pulse schemes. In particular, the threshold tunability via electrostatic gating enables the efficient in situ alteration of the probabilistic switching activation (PAct ) from 0 to 1.0, and can even modulate the degree of the PAct change. A drop-connected algorithm based on the PAct is constructed and used to successfully classify the shapes of several fashion items. The suggested approach can decrease the learning energy by up to ≈2,116 times relative to that of the conventional all-to-all connected network while exhibiting a high recognition accuracy of ≈93 %.

Controllable SiOx Nanorod Memristive Neuron for Probabilistic Bayesian Inference.

Modern artificial neural network technology using a deterministic computing framework is faced with a critical challenge in dealing with massive data that are largely unstructured and ambiguous. This challenge demands the advances of an elementary physical device for tackling these uncertainties. Here, we designed and fabricated a SiOx nanorod memristive device by employing the glancing angle deposition (GLAD) technique, suggesting a controllable stochastic artificial neuron that can mimic the fundamental integrate-and-fire signaling and stochastic dynamics of a biological neuron. The nanorod structure provides the random distribution of multiple nanopores all across the active area, capable of forming a multitude of Si filaments at many SiOx nanorod edges after the electromigration process, leading to a stochastic switching event with very high dynamic range (≈5.15 × 1010 ) and low energy (≈4.06 pJ). Different probabilistic activation (ProbAct) functions in a sigmoid form are implemented, showing its controllability with low variation by manufacturing and electrical programming schemes. Furthermore, as an application prospect, based on the suggested memristive neuron, we demonstrated the self-resting neural operation with the local circuit configuration and revealed probabilistic Bayesian inferences for genetic regulatory networks with low normalized mean squared errors (≈2.41 × 10-2 ) and its robustness to the ProbAct variation.

Tailoring the Interfacial Band Offset by the Molecular Dipole Orientation for a Molecular Heterojunction Selector.

Understanding and designing interfacial band alignment in a molecular heterojunction provides a foundation for realizing its desirable electronic functionality. In this study, a tailored molecular heterojunction selector is implemented by controlling its interfacial band offset between the molecular self-assembled monolayer with opposite dipole orientations and the 2D semiconductor (1L -MoS2 or 1L -WSe2 ). The molecular dipole moment direction determines the direction of the band bending of the 2D semiconductors, affecting the dominant transport pathways upon voltage application. Notably, in the molecular heterostructure with 1L -WSe2 , the opposite rectification direction is observed depending on the molecular dipole moment direction, which does not hold for the case with 1L -MoS2 . In addition, the nonlinearity of the molecular heterojunction selector can be significantly affected by the molecular dipole moment direction, type of 2D semiconductor, and metal work function. According to the choice of these heterojunction constituents, the nonlinearity is widely tuned from 1.0 × 101 to 3.6 × 104 for the read voltage scheme and from 0.4 × 101 to 2.0 × 105 for the half-read voltage scheme, which can be scaled up to an ≈482 Gbit crossbar array.

Reduced virulence of the MARTX toxin increases the persistence of outbreak-associated Vibrio vulnificus in host reservoirs.

Opportunistic bacteria strategically dampen their virulence to allow them to survive and propagate in hosts. However, the molecular mechanisms underlying virulence control are not clearly understood. Here, we found that the opportunistic pathogen Vibrio vulnificus biotype 3, which caused an outbreak of severe wound and intestinal infections associated with farmed tilapia, secretes significantly less virulent multifunctional autoprocessing repeats-in-toxin (MARTX) toxin, which is the most critical virulence factor in other clinical Vibrio strains. The biotype 3 MARTX toxin contains a cysteine protease domain (CPD) evolutionarily retaining a unique autocleavage site and a distinct ß-flap region. CPD autoproteolytic activity is attenuated following its autocleavage because of the ß-flap region. This ß-flap blocks the active site, disabling further autoproteolytic processing and release of the modularly structured effector domains within the toxin. Expression of this altered CPD consequently results in attenuated release of effectors by the toxin and significantly reduces the virulence of V. vulnificus biotype 3 in cells and in mice. Bioinformatic analysis revealed that this virulence mechanism is shared in all biotype 3 strains. Thus, these data provide new insights into the mechanisms by which opportunistic bacteria persist in an environmental reservoir, prolonging the potential to cause outbreaks.

Run-off election-based decision method for the training and inference process in an artificial neural network.

Generally, the decision rule for classifying unstructured data in an artificial neural network system depends on the sequence results of an activation function determined by vector-matrix multiplication between the input bias signal and the analog synaptic weight quantity of each node in a matrix array. Although a sequence-based decision rule can efficiently extract a common feature in a large data set in a short time, it can occasionally fail to classify similar species because it does not intrinsically consider other quantitative configurations of the activation function that affect the synaptic weight update. In this work, we implemented a simple run-off election-based decision rule via an additional filter evaluation to mitigate the confusion from proximity of output activation functions, enabling the improved training and inference performance of artificial neural network system. Using the filter evaluation selected via the difference among common features of classified images, the recognition accuracy achieved for three types of shoe image data sets reached ~ 82.03%, outperforming the maximum accuracy of ~ 79.23% obtained via the sequence-based decision rule in a fully connected single layer network. This training algorithm with an independent filter can precisely supply the output class in the decision step of the fully connected network.

Emerging Memristive Artificial Synapses and Neurons for Energy-Efficient Neuromorphic Computing.

Memristors have recently attracted significant interest due to their applicability as promising building blocks of neuromorphic computing and electronic systems. The dynamic reconfiguration of memristors, which is based on the history of applied electrical stimuli, can mimic both essential analog synaptic and neuronal functionalities. These can be utilized as the node and terminal devices in an artificial neural network. Consequently, the ability to understand, control, and utilize fundamental switching principles and various types of device architectures of the memristor is necessary for achieving memristor-based neuromorphic hardware systems. Herein, a wide range of memristors and memristive-related devices for artificial synapses and neurons is highlighted. The device structures, switching principles, and the applications of essential synaptic and neuronal functionalities are sequentially presented. Moreover, recent advances in memristive artificial neural networks and their hardware implementations are introduced along with an overview of the various learning algorithms. Finally, the main challenges of the memristive synapses and neurons toward high-performance and energy-efficient neuromorphic computing are briefly discussed. This progress report aims to be an insightful guide for the research on memristors and neuromorphic-based computing.

One-dimensional organic artificial multi-synapses enabling electronic textile neural network for wearable neuromorphic applications.

One-dimensional (1D) devices are becoming the most desirable format for wearable electronic technology because they can be easily woven into electronic (e-) textile(s) with versatile functional units while maintaining their inherent features under mechanical stress. In this study, we designed 1D fiber-shaped multi-synapses comprising ferroelectric organic transistors fabricated on a 100-µm Ag wire and used them as multisynaptic channels in an e-textile neural network for wearable neuromorphic applications. The device mimics diverse synaptic functions with excellent reliability even under 6000 repeated input stimuli and mechanical bending stress. Various NOR-type textile arrays are formed simply by cross-pointing 1D synapses with Ag wires, where each output from individual synapse can be integrated and propagated without undesired leakage. Notably, the 1D multi-synapses achieved up to ~90 and ~70% recognition accuracy for MNIST and electrocardiogram patterns, respectively, even in a single-layer neural network, and almost maintained regardless of the bending conditions.

MARTX Toxin-Stimulated Interplay between Human Cells and Vibrio vulnificus.

To understand toxin-stimulated host-pathogen interactions, we performed dual-transcriptome sequencing experiments using human epithelial (HT-29) and differentiated THP-1 (dTHP-1) immune cells infected with the sepsis-causing pathogen Vibrio vulnificus (either the wild-type [WT] pathogen or a multifunctional-autoprocessing repeats-in-toxin [MARTX] toxin-deficient strain). Gene set enrichment analyses revealed MARTX toxin-dependent responses, including negative regulation of extracellular related kinase 1 (ERK1) and ERK2 (ERK1/2) signaling and cell cycle regulation in HT-29 and dTHP-1 cells, respectively. Further analysis of the expression of immune-related genes suggested that the MARTX toxin dampens immune responses in gut epithelial cells but accelerates inflammation and nuclear factor κB (NF-κB) signaling in immune cells. With respect to the pathogen, siderophore biosynthesis genes were significantly more highly expressed in WT V. vulnificus than in the MARTX toxin-deficient mutant upon infection of dTHP-1 cells. Consistent with these results, iron homeostasis genes that limit iron levels for invading pathogens were overexpressed in WT V. vulnificus-infected dTHP-1 cells. Taken together, these results suggest that MARTX toxin regulates host inflammatory responses during V. vulnificus infection while also countering host defense mechanisms such as iron limitation.IMPORTANCEV. vulnificus is an opportunistic human pathogen that can cause life-threatening sepsis in immunocompromised patients via seafood poisoning or wound infection. Among the toxic substances produced by this pathogen, the MARTX toxin greatly contributes to disease progression by promoting the dysfunction and death of host cells, which allows the bacteria to disseminate and colonize the host. In response to this, host cells mount a counterattack against the invaders by upregulating various defense genes. In this study, the gene expression profiles of both host cells and V. vulnificus were analyzed by RNA sequencing to gain a comprehensive understanding of host-pathogen interactions. Our results suggest that V. vulnificus uses the MARTX toxin to subvert host cell immune responses as well as to oppose host counterattacks such as iron limitation.

Makes caterpillars floppy-like effector-containing MARTX toxins require host ADP-ribosylation factor (ARF) proteins for systemic pathogenicity.

Upon invading target cells, multifunctional autoprocessing repeats-in-toxin (MARTX) toxins secreted by bacterial pathogens release their disease-related modularly structured effector domains. However, it is unclear how a diverse repertoire of effector domains within these toxins are processed and activated. Here, we report that Makes caterpillars floppy-like effector (MCF)-containing MARTX toxins require ubiquitous ADP-ribosylation factor (ARF) proteins for processing and activation of intermediate effector modules, which localize in different subcellular compartments following limited processing of holo effector modules by the internal cysteine protease. Effector domains structured tandemly with MCF in intermediate modules become disengaged and fully activated by MCF, which aggressively interacts with ARF proteins present at the same location as intermediate modules and is converted allosterically into a catalytically competent protease. MCF-mediated effector processing leads ultimately to severe virulence in mice via an MCF-mediated ARF switching mechanism across subcellular compartments. This work provides insight into how bacteria take advantage of host systems to induce systemic pathogenicity.

Hepatitis B virus reactivation after radiotherapy for hepatocellular carcinoma and efficacy of antiviral treatment: A multicenter study.

Convincing data that support routine use of preventive therapy against hepatitis B virus (HBV) reactivation in radiotherapy (RT) for hepatocellular carcinoma (HCC) are lacking. The aim of this study was to investigate the incidence, clinical significance, and risk factors of HBV reactivation after RT. Medical records of 133 HBsAg (+) HCC patients who received radiotherapy from March 2009 to February 2016 were reviewed. Patients were divided into two groups: 1) non-antiviral group, those who did not receive antiviral therapy before RT (n = 27); and antiviral group (those who underwent antiviral therapy before RT) (n = 106). Factors related to HBV reactivation in HCC patients were evaluated. 17 (12.7%) of 133 patients developed HBV reactivation after RT. Patients in the antiviral group had significantly lower rates of HBV reactivation than those in the non-antiviral group (7.5% vs. 33.3%, p<0.001). HBV related hepatitis was also lower in the antiviral group (3.8% vs. 14.8%, p = 0.031). In multivariate analysis, absence of antiviral treatment (OR: 8.339, 95% CI: 2.532-27.470, p<0.001) and combined treatment of RT with transarterial chemoembolizatoin (TACE) (OR: 5.313, 95% CI: 1.548-18.232, p = 0.008) were risk factors for HBV reactivation. HBV reactivation can occur after radiotherapy. Combination treatment of RT with TACE and non-antiviral treatment are major risk factors for HBV reactivation during or after RT. Therefore, preventive antiviral therapy should be recommended for patients with HCC who are scheduled to receive RT.

Controllable Switching Filaments Prepared via Tunable and Well-Defined Single Truncated Conical Nanopore Structures for Fast and Scalable SiOx Memory.

The controllability of switching conductive filaments is one of the central issues in the development of reliable metal-oxide resistive memory because the random dynamic nature and formation of the filaments pose an obstacle to desirable switching performance. Here, we introduce a simple and novel approach to control and form a single silicon nanocrystal (Si-NC) filament for use in SiOx memory devices. The filament is formed with a confined vertical nanoscale gap by using a well-defined single vertical truncated conical nanopore (StcNP) structure. The physical dimensions of the Si-NC filaments such as number, size, and length, which have a significant influence on the switching properties, can be simply engineered by the breakdown of an Au wire through different StcNP structures. In particular, we demonstrate that the designed SiOx memory junction with a StcNP of pore depth of ∼75 nm and a bottom diameter of ∼10 nm exhibited a switching speed of up to 6 ns for both set and reset process, significantly faster than reported SiOx memory devices. The device also exhibited a high ON-OFF ratio, multistate storage ability, acceptable endurance, and retention stability. The influence of the physical dimensions of the StcNP on the switching features is discussed based on the simulated temperature profiles of the Au wire and the nanogap size generated inside the StcNP structure during electromigration.

Bayesian inference of the flow resistivity of a sound absorber and the room's influence on the Sabine absorption coefficients.

A Bayesian analysis is applied to determine the flow resistivity of a porous sample and the influence of the test chamber based on measured Sabine absorption coefficient data. The Sabine absorption coefficient measured in a reverberation chamber according to ISO 354 is influenced by the test chamber significantly, whereas the flow resistivity is a rather reproducible material property, from which the absorptive characteristics can be calculated through reliable models. Using Sabine absorption coefficients measured in 13 European reverberation chambers, the maximum a posteriori and the uncertainty of the flow resistivity and the test chamber's influence are estimated. Inclusion of more than one chamber's absorption data helps the flow resistivity converge towards a reliable value with a standard deviation below 17%.