Hydra: Multi-head low-rank adaptation for parameter efficient fine-tuning.

The recent surge in large-scale foundation models has spurred the development of efficient methods for adapting these models to various downstream tasks. Low-rank adaptation methods, such as LoRA, have gained significant attention due to their outstanding parameter efficiency and no additional inference latency. This paper investigates a more general form of adapter module based on the analysis that parallel and sequential adaptation branches learn novel and general features during fine-tuning, respectively. The proposed method, named Hydra, combines parallel and sequential branch to integrate capabilities, which is more expressive than existing single branch methods and enables the exploration of a broader range of optimal points in the fine-tuning process. In addition, the proposed method explicitly leverages the pre-trained weights by performing a linear combination of the pre-trained features. It allows the learned features to have better generalization performance across diverse downstream tasks. Furthermore, we perform a comprehensive analysis of the characteristics of each adaptation branch with empirical evidence. Through an extensive range of experiments, we substantiate the efficiency and demonstrate the superior performance of Hydra. This comprehensive evaluation underscores the potential impact and effectiveness of Hydra in a variety of applications. The source code of this work is publicly opened on https://github.com/extremebird/Hydra.

Field-Free Spin-Orbit Torque Magnetization Switching in a Single-Phase Ferromagnetic and Spin Hall Oxide.

Current-induced spin-orbit torque (SOT) offers substantial promise for the development of low-power, nonvolatile magnetic memory. Recently, a single-phase material concurrently exhibiting magnetism and the spin Hall effect has emerged as a scientifically and technologically interesting platform for realizing efficient and compact SOT systems. Here, we demonstrate external-magnetic-field-free switching of perpendicular magnetization in a single-phase ferromagnetic and spin Hall oxide SrRuO3. We delicately altered the local lattices of the top and bottom surface layers of SrRuO3, while retaining a quasi-homogeneous, single-crystalline nature of the SrRuO3 bulk. This leads to unbalanced spin Hall effects between the top and bottom layers, enabling net SOT performance within single-layer ferromagnetic SrRuO3. Notably, our SrRuO3 exhibits the highest SOT efficiency and lowest power consumption among all known single-layer systems under field-free conditions. Our method of artificially manipulating the local atomic structures will pave the way for advances in spin-orbitronics and the exploration of new SOT materials.

Willingness to take responsibility: Self-sacrifice versus sacrificing others in takeover decisions during autonomous driving.

In Level-3 autonomous driving, drivers are required to take over in an emergency upon receiving a request from an autonomous vehicle (AV). However, before the deadline for the takeover request expires, drivers are not considered fully responsible for the accident, which may make them hesitant to assume control and take on full liability before the time runs out. Therefore, to prevent problems caused by late takeover, it is important to know which factors influence a driver's willingness to take over in an emergency. To address this issue, we recruited 250 participants each for both video-based and text-based surveys to investigate the takeover decision in a dilemmatic situation that can endanger the driver, with the AV either sacrificing a group of pedestrians or the driver if the participants do not intervene. The results showed that 88.2% of respondents chose to take over when the AV intended to sacrifice the driver, while only 59.4% wanted to take over when the pedestrians would be sacrificed. Additionally, when the AV's chosen path matched the participant's intention, 77.4% chose to take over when the car intended to sacrifice the driver compared with only 34.3% when the pedestrians would be sacrificed. Furthermore, other factors such as sex, driving experience, and driving preferences partially influenced takeover decisions; however, they had a smaller effect than the situational context. Overall, our findings show that regardless of the driving intention of an AV, informing drivers that their safety is at risk can enhance their willingness to take over control of an AV in critical situations.

Usefulness of sectional images in dural AVF for the interpretation of venous anatomy.

Knowledge of the venous anatomy is essential for appropriately treating dural arteriovenous fistulas (AVFs). It is challenging to determine the overall venous structure despite performing selective angiography for dural AVFs with feeder from multiple selected arteries. This is because only a part of the veins can be observed through the shunt in the selected artery. Therefore, after performing selective angiography of all vessels to understand the approximate venous anatomy, the venous anatomy can be easily understood by closely examining the source image of computed tomographic angiography or magnetic resonance angiography. Through this, it is possible to specify the vein that is to be blocked (target embolization), thereby avoiding extensive blocking of the vein and avoiding various complications. In the case of dural AVF with feeder from single selected artery, if the multiplanar reconstruction image of the three-dimensional rotational computed tomography obtained by performing angiography is analyzed thoroughly, a shunted pouch can be identified. If embolization is performed by targeting this area, unnecessary sinus total packing can be avoided.

Establishment and application of bioassay- and molecular marker-based methods for monitoring fluvalinate resistance of Varroa mites.

The Varroa mite, Varroa destructor, is an ectoparasite that infests honey bees. The extensive use of acaricides, including fluvalinate, has led to the emergence of resistance in Varroa mite populations worldwide. This study's objective is to monitor fluvalinate resistance in field populations of Varroa mites in Korea through both bioassay-based and molecular marker-based methods. To achieve this, a residual contact vial (RCV) bioassay was established for on-site resistance monitoring. A diagnostic dose of 200 ppm was determined based on the bioassay using a putative susceptible population. In the RCV bioassay, early mortality evaluation was effective for accurately discriminating mites with the knockdown resistance (kdr) genotype, while late evaluation was useful for distinguishing mites with additional resistance factors. The RCV bioassay of 14 field mite populations collected in 2021 indicated potential resistance development in four populations. As an alternative approach, quantitative sequencing was employed to assess the frequency of the L925I/M mutation in the voltage-gated sodium channel (VGSC), associated with fluvalinate kdr trait. While the mutation was absent in 2020 Varroa mite populations, it emerged in 2021, increased in frequency in 2022, and became nearly widespread across the country by 2023. This recent emergence and rapid spread of fluvalinate resistance within a span of three years demonstrate the Varroa mite's significant potential for developing resistance. This situation further underscores the urgent need to replace fluvalinate with alternative acaricides. A few novel VGSC mutations potentially involved in resistance were identified. Potential factors driving the rapid expansion of resistance were further discussed.

Interpretation of SNP combination effects on schizophrenia etiology based on stepwise deep learning with multi-precision data.

Schizophrenia genome-wide association studies (GWAS) have reported many genomic risk loci, but it is unclear how they affect schizophrenia susceptibility through interactions of multiple SNPs. We propose a stepwise deep learning technique with multi-precision data (SLEM) to explore the SNP combination effects on schizophrenia through intermediate molecular and cellular functions. The SLEM technique utilizes two levels of precision data for learning. It constructs initial backbone networks with more precise but small amount of multilevel assay data. Then, it learns strengths of intermediate interactions with the less precise but massive amount of GWAS data. The learned networks facilitate identifying effective SNP interactions from the intractably large space of all possible SNP combinations. We have shown that the extracted SNP combinations show higher accuracy than any single SNPs and preserve the accuracy in an independent dataset. The learned networks also provide interpretations of molecular and cellular interactions of SNP combinations toward schizophrenia etiology.

Acetylcholine titre regulation by non-neuronal acetylcholinesterase 1 and its putative roles in honey bee physiology.

Similar to other insects, honey bees have two acetylcholinesterases (AChEs), AmAChE1 and AmAChE2. The primary catalytic enzyme for acetylcholine (ACh) hydrolysis in synapses is AmAChE2, which is predominantly expressed in neuronal tissues, whereas AmAChE1 is expressed in both neuronal and non-neuronal tissues, with limited catalytic activity. Unlike constitutively expressed AmAChE2, AmAChE1 expression is induced under stressful conditions such as heat shock and brood rearing suppression, but its role in regulating ACh titre remains unclear. In this paper, to elucidate the role of AmAChE1, the expression of AmAChE1 was suppressed via RNA interference (RNAi) in AmAChE1-induced worker bees. The ACh titre measurement following RNAi revealed that the expression of AmAChE1 downregulated the overall ACh titre in all tissues examined without altering AmAChE2 expression. Transcriptome analysis showed that AmAChE1 knockdown upregulated protein biosynthesis, cell respiration, and thermogenesis in the head. These findings suggest that AmAChE1 is involved in decreasing neuronal activity, enhancing energy conservation, and potentially extending longevity under stressful conditions via ACh titre regulation.

Endovascular Management of Iatrogenic Vertebral Artery Pseudoaneurysm: A Case Report.

Iatrogenic injury of the vertebral artery during cervical spine surgery though uncommon is critical. With advances in interventional endovascular techniques, the therapeutic approach for vertebral artery injuries has changed. Nonetheless, an established strategy for their management is lacking. We report a case of pseudoaneurysm due to vertebral artery injury, during cervical spine surgery for a tumor, that was treated successfully with endovascular coiling in a plug-and-patch fashion after triple stenting failed.

Feasibility of Prompt Lumbar Drainage in Patients with Aneurysmal Subarachnoid Hemorrhage.

BACKGROUND: Numerous studies have shown that continuous lumbar drainage (LD) reduces spontaneous subarachnoid hemorrhage (SAH)-related complications, decreasing the incidence of cerebral vasospasm, delayed cerebral ischemia , and hydrocephalus in patients treated with coiling or clipping, but performing LD before securing the aneurysm is still controversial. Our hospital has been implementing prompt LD for several years, and we present the results in this paper. METHODS: Between January 2014 and December 2020, a total of 438 patients with SAH were included in this retrospective study. The indication for prompt LD was aneurysmal SAH of modified Fisher grade III or higher without dense intraventricular hemorrhage with obstructive hydrocephalus requiring extraventricular drainage or large intracranial hemorrhage requiring immediate decompression. Prompt LD was performed for 229 patients with SAH, and the control group included 209 patients. We compared in-hospital mortality and vasospasm or hydrocephalus occurrence and procedure-related complications between the two groups. RESULTS: The in-hospital mortality rate was 7.4% for patients with prompt LD and 14.4% for patients without LD, and the difference was significant (P = 0.019). Vasospasm occurred in 10% of patients with prompt LD and 16.7% of controls (P = 0.039). Hydrocephalus requiring extraventricular drainage occurred in 10.9% of the LD group and 28.7% of the control group (P < 0.001). Rebleeding occurrence was 3.1% in the prompt LD group and 5.7% in the non-LD group (P = 0.168). Cerebrospinal fluid infection occurred in 0.4% of the prompt LD group and 1.4% of controls(P = 0.272). CONCLUSIONS: Prompt LD is a feasible option for treating patients with selective aneurysmal SAH.

Demonstration of programmable light intensity of a micro-LED with a Hf-based ferroelectric ITZO TFT for Mura-free displays.

We demonstrate the programmable light intensity of a micro-LED by compensating threshold voltage variability of thin-film transistors (TFTs) by introducing a non-volatile programmable ferroelectric material, HfZrO2 (HZO) into the gate stack of the TFT. We fabricated an amorphous ITZO TFT, ferroelectric TFTs (FeTFTs), and micro-LEDs and verified the feasibility of our proposed current-driving active matrix circuit. Importantly, we successfully present the programmed multi-level lighting of the micro-LED, utilizing partial polarization switching in the a-ITZO FeTFT. We expect that this approach will be highly promising for the next-generation display technology, replacing complicated threshold voltage compensation circuits with a simple a-ITZO FeTFT.

Imaging features of hyperacute intracerebral hemorrhage on multiple MRI sequences within 1 minute from onset during MRI examination: A case report.

RATIONALE: Acute stroke requires accurate imaging to ensure appropriate treatment decisions and favorable clinical outcomes. Computed tomography has long been used as an exclusive imaging technique to assess intracerebral hemorrhage, owing to its rapid scanning time and widespread availability. Several recent studies have reported the reliable detection of hyperacute hemorrhage using magnetic resonance imaging (MRI). PATIENT CONCERNS: An 88-year-old woman with a history of hypertension presented with mild, acute dysarthria. The National Institutes of Health Stroke Scale score was 1. DIAGNOSES: Non contrast head computed tomography revealed the absence of acute cerebral hemorrhage. The patient underwent magnetic resonance, revealed hyperacute intracerebral hemorrhage within a few minutes of its occurrence on multiple MRI sequences. INTERVENTIONS AND OUTCOMES: In this patient, hemorrhage developed during MRI for acute ischemic stroke. Hemorrhage was initially misdiagnosed, and inappropriate treatment severely affected the patient's health. LESSONS: Clinicians in the Department of Neurological Emergency should be familiar with imaging findings of hyperacute hemorrhage on multiple MRI sequences.

Pretreatment methods for monitoring microplastics in soil and freshwater sediment samples: A comprehensive review.

This paper reviews the currently used pretreatment methods for microplastics (MPs) analysis in soil and freshwater sediments, primarily sample processing, pretreatment, and characterization methods for MPs analysis. In addition, analytical tools (e.g., lab instruments), MPs characteristics, and MPs quantity, are included in this review. Prior to pretreatment, soil and sediment samples are typically processed using sieving and drying methods, and a sample quantity of <50 g was mostly used for the pretreatment. Density separation was commonly performed before organic matter removal. Sodium chloride (NaCl) and zinc chloride (ZnCl2) were most often used for density separation, and hydrogen peroxide (H2O2) oxidation was most frequently used to remove organic matter. Although advantages of each pretreatment method have been investigated, it is still challenging to determine a universal pretreatment method due to sample variability (e.g., sample characteristics). Furthermore, it is highly required to establish standard pretreatment methods that can be used for various environmental matrices, including air, water, and wastes as well as soil and sediment.

Dielectric-Engineered High-Speed, Low-Power, Highly Reliable Charge Trap Flash-Based Synaptic Device for Neuromorphic Computing beyond Inference.

The coming of the big-data era brought a need for power-efficient computing that cannot be realized in the Von Neumann architecture. Neuromorphic computing which is motivated by the human brain can greatly reduce power consumption through matrix multiplication, and a device that mimics a human synapse plays an important role. However, many synaptic devices suffer from limited linearity and symmetry without using incremental step pulse programming (ISPP). In this work, we demonstrated a charge-trap flash (CTF)-based synaptic transistor using trap-level engineered Al2O3/Ta2O5/Al2O3 gate stack for successful neuromorphic computing. This novel gate stack provided precise control of the conductance with more than 6 bits. We chose the appropriate bias for highly linear and symmetric modulation of conductance and realized it with very short (25 ns) identical pulses at low voltage, resulting in low power consumption and high reliability. Finally, we achieved high learning accuracy in the training of 60000 MNIST images.

Short communication: Screening of proof-of-concept mutations of honey bee acetylcholinesterase 2 conferring resistance to organophosphorus and carbamate insecticides.

The introduction of pesticide resistance-inducing mutations into target genes would in theory protect honey bees from the hazardous effects of pesticides. In this paper, to screen amino acid substitutions conferring resistance to organophosphorus and carbamate insecticides, honey bee acetylcholinesterase 2 (AmAChE2) variants with several mutations (V260L, A316S, G342A, G342V, F407Y, and G342V/F407Y) were generated and expressed in vitro using a baculovirus system. The inhibition constants of recombinant native and mutated AmAChE2s against six pesticides were measured. As a result, the A316S mutation was shown to induce high resistance without a catalytic efficiency change.

Overexpression of transmembrane TNFα in brain endothelial cells induces schizophrenia-relevant behaviors.

Upregulation of genes and coexpression networks related to immune function and inflammation have been repeatedly reported in the brain of individuals with schizophrenia. However, a causal relationship between the abnormal immune/inflammation-related gene expression and schizophrenia has not been determined. We conducted co-expression networks using publicly available RNA-seq data from prefrontal cortex (PFC) and hippocampus (HP) of 64 individuals with schizophrenia and 64 unaffected controls from the SMRI tissue collections. We identified proinflammatory cytokine, transmembrane tumor necrosis factor-α (tmTNFα), as a potential regulator in the module of co-expressed genes that we find related to the immune/inflammation response in endothelial cells (ECs) and/or microglia of the brain of individuals with schizophrenia. The immune/inflammation-related modules associated with schizophrenia and the TNF signaling pathway that regulate the network were replicated in an independent cohort of brain samples from 68 individuals with schizophrenia and 135 unaffected controls. To investigate the association between the overexpression of tmTNFα in brain ECs and schizophrenia-like behaviors, we induced short-term overexpression of the uncleavable form of (uc)-tmTNFα in ECs of mouse brain for 7 weeks. We found schizophrenia-relevant behavioral deficits in these mice, including cognitive impairment, abnormal sensorimotor gating, and sensitization to methamphetamine (METH) induced locomotor activity and METH-induced neurotransmitter levels. These uc-tmTNFα effects were mediated by TNF receptor2 (TNFR2) and induced activation of TNFR2 signaling in astrocytes and neurons. A neuronal module including neurotransmitter signaling pathways was down-regulated in the brain of mice by the short-term overexpression of the gene, while an immune/inflammation-related module was up-regulated in the brain of mice after long-term expression of 22 weeks. Our results indicate that tmTNFα may play a direct role in regulating neurotransmitter signaling pathways that contribute to the clinical features of schizophrenia.

Molecular and kinetic properties of three acetylcholinesterases in the Varroa mite, Varroa destructor.

The Varroa mite, Varroa destructor, poses one of the most serious threats to honey bees worldwide. Although coumaphos, an anticholinesterase pesticide, is widely used for varroa mite control, little information is available on the properties of Varroa mite acetylcholinesterases (VdAChEs). In this study, three putative VdAChEs were annotated and named VdAChE1, VdAChE2, and VdAChE3. All VdAChEs possessed most of the functionally important signature domains, suggesting that they are catalytically active. Phylogenetic analysis revealed that VdAChE1 was clustered into a clade containing most arthropod AChE1s, whereas VdAChE2 and VdAChE3 formed a unique clade with other arachnid AChEs. VdAChE1 was determined to be membrane-anchored, but both VdAChE2 and VdAChE3 are soluble, as judged by electrophoresis in conjunction with western blotting. Tissue-specific transcription profiling revealed that VdAChE1 was most predominantly expressed in the synganglion. In contrast, VdAChE2 was most predominantly expressed in the legs and cuticle. VdAChE3 showed negligible expression levels in all the tissues examined. In a kinetic analysis using recombinant VdAChEs, VdAChE1 exhibited the highest catalytic efficiency, followed by VdAChE2 and VdAChE3. Inhibition experiments revealed that VdAChE1 was most sensitive to all tested inhibitors. Taken together, VdAChE1 appears to be the major synaptic enzyme with a more toxicological relevance, whereas VdAChE2 is involved in other noncatalytic functions, including chemical defense against xenobiotics. Current findings contribute to a more detailed understanding of the evolutionary and functional traits of VdAChEs and to the design of novel anticholinesterase varroacides.

High-sensitivity waveguide-integrated bolometer based on free-carrier absorption for Si photonic sensors.

Conventional photon detectors necessarily face critical challenges regarding strong wavelength-selective response and narrow spectral bandwidth, which are undesirable for spectroscopic applications requiring a wide spectral range. With this perspective, herein, we overcome these challenges through a free-carrier absorption-based waveguide-integrated bolometer for infrared spectroscopic sensors on a silicon-on-insulator (SOI) platform featuring a spectrally flat response at near-infrared (NIR) range (1520-1620 nm). An in-depth thermal analysis was conducted with a systematic investigation of geometry dependence on the detectors. We achieved great performances: temperature coefficient of resistance (TCR) of -3.786%/K and sensitivity of -26.75%/mW with a low wavelength dependency, which are record-high values among reported waveguide bolometers so far, to our knowledge. In addition, a clear on-off response with the rise/fall time of 24.2/29.2 µs and a 3-dB roll-off frequency of ∼22 kHz were obtained, sufficient for a wide range of sensing applications. Together with the possibility of expanding an operation range to the mid-infrared (MIR) band, as well as simplicity in the detector architecture, our work here presents a novel strategy for integrated photodetectors covering NIR to MIR at room temperature for the development of the future silicon photonic sensors with ultrawide spectral bandwidth.

Atomic-scale mapper for superlattice photodetectors analysis.

In this work, a new Python-based tool for atomic-scale mapping of high-angle annular dark-field (HAADF) and annular bright-field (ABF) scanning transmission electron microscopy (STEM) images using the Z-contrast method is introduced, aimed to help in the analysis of superlattice layers' composition, and in the determination of material of interfaces. The operation principle of the program, as well as specific examples of use, are explained in many details. Good customization flexibility using the user-friendly graphical user interface (GUI), allows the processing of a wide range of images, demonstrating a decent accuracy of coordinates extraction and performance.

CT and MRI Features of Middle Ear Fibrous Hamartoma of Infancy: A Case Report.

Fibrous hamartoma of infancy in the middle ear is extremely rare. We report the case of a 26-month-old male patient who presented with a mass in the left middle ear. A temporal bone CT scan showed complete opacification of the left middle ear and mastoid air cells without ossicular erosion. On MRI, the mass revealed heterogeneous signal intensities indicative of fat and fibrous components. A definitive diagnosis was made postoperatively based on the histological results. Although rare, fibrous hamartoma of infancy should be considered as a differential diagnosis of a middle ear mass during childhood.

Flexoelectricity-Driven Mechanical Switching of Polarization in Metastable Ferroelectrics.

Flexoelectricity-based mechanical switching of ferroelectric polarization has recently emerged as a fascinating alternative to conventional polarization switching using electric fields. Here, we demonstrate hyperefficient mechanical switching of polarization exploiting metastable ferroelectricity that inherently holds a unique mechanical response. We theoretically predict that mechanical forces markedly reduce the coercivity of metastable ferroelectricity, thus greatly bolstering flexoelectricity-driven mechanical polarization switching. As predicted, we experimentally confirm the mechanical polarization switching via an unusually low mechanical force (100 nN) in metastable ferroelectric CaTiO_{3}. Furthermore, the use of low mechanical forces narrows the width of mechanically writable nanodomains to sub-10 nm, suggesting an ultrahigh data storage density of ≥1 Tbit cm^{-2}. This Letter sheds light on the mechanical switching of ferroelectric polarization as a viable key element for next-generation efficient nanoelectronics and nanoelectromechanics.