Multimode microdimer robot for crossing tissue morphological barrier

Multimodal Transformers achieve superior performance in multimodal learning tasks. However, the quadratic complexity of the self-attention mechanism in Transformers limits their deployment in low-resource devices and makes their inference and training computationally expensive. We propose multimodal Sparse Phased Transformer (SPT) to alleviate the problem of self-attention complexity and memory footprint. SPT uses a sampling function to generate a sparse attention matrix and compress a long sequence to a shorter sequence of hidden states. SPT concurrently captures interactions between the hidden states of different modalities at every layer. To further improve the efficiency of our method, we use Layer-wise parameter sharing and Factorized Co-Attention that share parameters between Cross Attention Blocks, with minimal impact on task performance. We evaluate our model with three sentiment analysis datasets and achieve comparable or superior performance compared with the existing methods, with a 90% reduction in the number of parameters. We conclude that (SPT) along with parameter sharing can capture multimodal interactions with reduced model size and improved sample efficiency.

M2TNet: Multi-modal multi-task Transformer network for ultra-short-term wind power multi-step forecasting

Under the dual constraints of strong industrial noise and small sample size, achieving high precision and reliability in pipeline leakage detection is a significant challenge. The root cause lies in the fact that traditional single-modal methods have a single perception dimension and insufficient anti-interference ability, making it difficult to effectively distinguish strong background noise with overlapping spectra, resulting in reduced measurement accuracy and reliability. To address this issue, this paper proposes an adaptive noise reduction detection method based on multi-modal Transformer and contrastive learning (MDT-CL). This paper uses pressure, vibration, and acoustic emission sensors to capture complementary signals, and synchronizes the collection of these signals through dynamic time warping technology. The proposed method comprises three key innovations. Firstly, we propose a multi-scale Transformer encoder based on physical principles, employing an asymmetric cross-modal attention mechanism to focus on pressure signals that are sensitive to leakage. We integrate vibration and acoustic features and fuse them at the feature level, fundamentally solving the heterogeneity and spatio-temporal mismatch problems of multi-source asynchronous sensor data. Secondly, we propose an embedded adaptive spectral normalization module, which is a dynamic frequency-domain filtering technology that collaboratively optimizes with the network. It intelligently suppresses noise-sensitive frequency bands and synchronously enhances leakage features, effectively overcoming the limitations of traditional static filtering. Thirdly, we propose a contrastive learning pre-training strategy for pipeline noise. By constructing a multi-modal noise enhancement library and feature decoupling mechanism, it significantly expands the inter-class distance between leakage and normal operating condition noise, enhancing the model’s generalization ability in complex noise conditions. Experimental results demonstrate that the proposed method achieves a leakage detection accuracy of 96.8%, representing a 1.6% improvement over the current state-of-the-art method.

Swimming microrobots that are energized by external magnetic fields exhibit a variety of intriguing collective behaviors, ranging from dynamic self-organization to coherent motion; however, achieving multiple, desired collective modes within one colloidal system to emulate high environmental adaptability and enhanced tasking capabilities of natural swarms is challenging. Here, we present a strategy that uses alternating magnetic fields to program hematite colloidal particles into liquid, chain, vortex, and ribbon-like microrobotic swarms and enables fast and reversible transformations between them. The chain is characterized by passing through confined narrow channels, and the herring school-like ribbon procession is capable of large-area synchronized manipulation, whereas the colony-like vortex can aggregate at a high density toward coordinated handling of heavy loads. Using the developed discrete particle simulation methods, we investigated generation mechanisms of these four swarms, as well as the "tank-treading" motion of the chain and vortex merging. In addition, the swarms can be programmed to steer in any direction with excellent maneuverability, and the vortex's chirality can be rapidly switched with high pattern stability. This reconfigurable microrobot swarm can provide versatile collective modes to address environmental variations or multitasking requirements; it has potential to investigate fundamentals in living systems and to serve as a functional bio-microrobot system for biomedicine.

Wireless communications at high-frequency bands with large antenna arrays face challenges in beam management, which can potentially be improved by multimodality sensing information from cameras, LiDAR, radar, and GPS. In this paper, we present a multimodal transformer deep learning framework for sensing-assisted beam prediction. We employ a convolutional neural network to extract the features from a sequence of images, point clouds, and radar raw data sampled over time. At each convolutional layer, we use transformer encoders to learn the hidden relations between feature tokens from different modalities and time instances over abstraction space and produce encoded vectors for the next-level feature extraction. We train the model on a combination of different modalities with supervised learning. We try to enhance the model over imbalanced data by utilizing focal loss and exponential moving average. We also evaluate data processing and augmentation techniques such as image enhancement, segmentation, background filtering, multimodal data flipping, radar signal transformation, and GPS angle calibration. Experimental results show that our solution trained on image and GPS data produces the best distance-based accuracy of predicted beams at 78.44%, with effective generalization to unseen day scenarios near 73% and night scenarios over 84%. This outperforms using other modalities and arbitrary data processing techniques, which demonstrates the effectiveness of transformers with feature fusion in performing radio beam prediction from images and GPS. Furthermore, our solution could be pretrained from large sequences of multimodality wireless data, on fine-tuning for multiple downstream radio network tasks.

SummaryIonotropic glutamate receptors (iGluRs) harbor two extracellular domains: the membrane-proximal ligand-binding domain (LBD) and the distal N-terminal domain (NTD). These are involved in signal sensing: the LBD binds L-glutamate, which activates the receptor channel. Ligand binding to the NTD modulates channel function in the NMDA receptor subfamily of iGluRs, which has not been observed for the AMPAR subfamily to date. Structural data suggest that AMPAR NTDs are packed into tight dimers and have lost their signaling potential. Here, we assess NTD dynamics from both subfamilies, using a variety of computational tools. We describe the conformational motions that underly NMDAR NTD allosteric signaling. Unexpectedly, AMPAR NTDs are capable of undergoing similar dynamics; although dimerization imposes restrictions, the two subfamilies sample similar, interconvertible conformational subspaces. Finally, we solve the crystal structure of AMPAR GluA4 NTD, and combined with molecular dynamics simulations, we characterize regions pivotal for an as-yet-unexplored dynamic spectrum of AMPAR NTDs.

In nonaqueous organic solvents, lipases can catalyze esterification reactions, which increase their application value. Yarrowia lipolytica Lipase (Lip2) possesses potential values in medicine and industrial production. In order to investigate its lid closure mechanism in methanol we performed molecular dynamics (MD) simulations of the open conformation of Lip2 in methanol and hexane, respectively. Simulation results indicated that Lip2 undergoes a greater conformational change in methanol. Principle component analysis showed Lip2 has "double-domain" and "torsion" motion modes in hexane and methanol. By analyzing B-factor and dynamical cross-correlation, region Ser274-Asn288, region Thr106-His126, and region Asp61-Asp67 were found to interact with the lid region (Thr88-Leu105). Furthermore, local restricted MD simulations showed that closure mechanism of Lip2 is "double-lid movement" which is also observed in Pseudomonas aeruginosa Lipase (PAL), and we detected two interaction propagation pathways in Lip2 driven by the interaction between Ser274-Asn288 and methanol.

Nowadays, deep-waters are the new frontier of the latest huge oil discovers. The most common technology for oil exploration in deep waters is based on the FPSO concept. Therefore, a key element for a successful exploration of deepwater oil fields is the accurate prediction of the unit motions in order to minimize the stop production due to excessive motions and maximum accelerations to which the process plant equipment and supports are subjected. The excessive motions of a FPSO can be caused not only due to rough weather but also due to mild weather at beam sea conditions. Beam sea conditions are faced by FPSOs in a spreading mooring configuration or even in single point mooring configuration but moored in regions where the swell sea is important, for instance, in Campos Basin offshore Brazil. In such cases, the accurate evaluation of roll motions is of prime importance. Moreover, in order to reach valuable results, a critical step is the proper evaluation of the different damping effects in particular viscous damping induced by the bilge keel, or those of risers. The state of the art assumption is that the roll damping is composed of two parts: the linear one and the quadratic one. The former accounts for the potential wave radiation and partly to some frictional effects, while the latter accounts for drag forces, flow separation, and other non-linear effects. In practice, the linear part can usually be neglected and the only remaining damping is the quadratic part. These fact strengths the role played by the non-potential damping in a reliable evaluation of the roll motion amplitude. In this paper, we will show the implementation of a practical procedure for the evaluation of the roll motions including the non-potential damping of a FPSO moored in a spreading configuration at Campos Basin and its validation through the comparison of the numerical results to the model tests. Introduction Compared to the other modes of motion, the evaluation of the roll motion appears to be much more complicated and the corresponding results more uncertain. The reasons for that are mainly related to the evaluation of the roll damping coefficient, which is usually dominated by the non-potential sources of damping. The usual procedure for calculation of the ship motions is the use of the so-called diffraction-radiation software based on the potential flow theory. The damping calculated by this kind of codes is linear and related to the wave making. As far as the linear theory is concerned the horizontal modes of motion (surge, sway, yaw) are usually out of resonance in the common sea conditions (periods of 5 to 25 seconds). In those cases the role of damping is not very important so that the potential flow theory evaluate correctly the (linear) amplitudes of these motions. The only modes of motion that may enter in the (linear) resonance are vertical modes (heave, pitch and roll) and for these motions the role of the damping is essential.

Flow instability analysis in U-tubes of steam generator under rolling movement

Computational Methods for Efficient Sampling of Protein Landscapes and Disclosing Allosteric Regions.

High precision approximation of the intermolecular potential is necessary for molecular dynamics (MD) simulations of the physical properties of molecular nitrogen (N2). Recently, Quantum Mechanical (QM) high precision calculations (coupled-cluster singles-and-doubles with a perturbative triples corrections [CCSD(T)]) were used to determine N2 - N2 intermolecular interaction. Using these data, the potential, covering all orientations of the pair of the nitrogen molecules, was obtained in the form invariant with respect of rotation of the reference system. The new N2-N2 potential remains in reasonably good agreement with the isotropically scaled potential function of van der Avoird et al, based on the results of Hartree–Fock calculations. Nitrogen equation of state, at pressures up to 30 GPa (300 kbar) and at temperatures from room temperature to 2000K, was obtained by MD simulations, employing the following potentials: the new CCSD(T) function, the nonscaled and scaled expression by van der Avoird et al., and four-atom interaction by Johnson et al. It was shown that the new potential gives considerably better agreement with the experimental data than in all other cases. MD simulations were also used to determine shear viscosity of nitrogen. It was shown that the influence on the potential variation on the MD obtained values of the viscosity is much stronger than on the equation of state, especially at higher densities. The results were compared with the experimentally measured values of the viscosity showing considerable difference between the date obtained from experiment and derived form MD simulations.

CDK2 can be used as an attractive target for development of efficient inhibitors curing multiple disease relating with CDK2. In this work, molecular dynamics (MD) simulations and binding free energy calculations were coupled to probe conformational changes of CDK2 due to inhibitor associations and binding mechanisms of inhibitors PM1, FMD and X64 to CDK2. The results suggest that the binding strength of FMD and X64 to CDK2 is stronger than that of PM1. Principal component (PC) analysis and cross-correlation map calculations based on the equilibrated MD trajectories demonstrate that the structural difference in inhibitors exerts important impact on motion modes and dynamics behavior of CDK2. Residue-based free energy decomposition method was adopted to estimate the inhibitor–residue spectrum. The results not only efficiently identify the hot interaction spot of inhibitors with CDK2 but also show that the hydrophobic rings R1, R2 and R3 as well as polar groups of three inhibitors play key roles in favorably binding of inhibitors to CDK2. This work is expected to contribute energetic basis and dynamics information to development of promising inhibitors toward CDK2.Communicated by Ramaswamy H. Sarma

HIV-1 protease (PR) is thought to be efficient targets of anti-AIDS drug design. Molecular dynamics (MD) simulations and multiple post-processing analysis technologies were applied to decipher molecular mechanism underlying binding of three drugs Lopinavir (LPV), Nelfinavir (NFV) and Atazanavir (ATV) to the PR. Binding free energies calculated by molecular mechanics generalized Born surface area (MM-GBSA) suggest that compensation between binding enthalpy and entropy plays a vital role in binding of drugs to PR. Dynamics analyses show that binding of LPV, NFV and ATV highly affects structural flexibility, motion modes and dynamics behaviour of the PR, especially for two flaps. Computational alanine scanning and interaction network analysis verify that although three drugs have structural difference, they share similar binding modes to the PR and common interaction clusters with the PR. The current findings also confirm that residues located interaction clusters, such as Asp25/Asp25ʹ, Gly27/Gly27ʹ, Ala28/Ala28ʹ, Asp29, Ile47/Ile47ʹ, Gly49/Gly49ʹ, Ile50/Ile50ʹ, Val82/Val82ʹ and Ile84/Ile84, can be used as efficient targets of clinically available inhibitors towards the PR.

Mutations V32I, I50V and I84V in the HIV-1 protease (PR) induce drug resistance towards drug amprenavir (APV). Multiple short molecular dynamics (MSMD) simulations and molecular mechanics generalized Born surface area (MM-GBSA) method were utilized to investigate drug-resistant mechanism of V32I, I50V and I84V towards APV. Dynamic information arising from MSMD simulations suggest that V32I, I50V and I84V highly affect structural flexibility, motion modes and conformational behaviours of two flaps in the PR. Binding free energies calculated by MM-GBSA method suggest that the decrease in binding enthalpy and the increase in binding entropy induced by mutations V32I, I50V and I84V are responsible for drug resistance of the mutated PRs on APV. The energetic contributions of separate residues on binding of APV to the PR show that V32I, I50V and I84V highly disturb the interactions of two flaps with APV and mostly drive the decrease in binding ability of APV to the PR. Thus, the conformational changes of two flaps in the PR caused by V32I, I50V and I84V play key roles in drug resistance of three mutated PR towards APV. This study can provide useful dynamics information for the design of potent inhibitors relieving drug resistance.

Event Abstract Back to Event Hemispheric Specialization for Processing Arithmetic in Adults Veronica Connaughton1*, Vicole Bothma2, Azhani Amiruddin2, Karen Clunies-Ross2, Noel French3 and Allison Fox4 1 Univeristy of Western Australia, School of Psychology, Australia 2 University of Western Australia, Australia 3 King Edward Memorial Hospital, Australia 4 University of Western Australia, School of Psychology, Australia A variety of current models of the cerebral circuits that underpin arithmetic calculation exist although research into cerebral lateralization of mathematical operations is in its initial stages. This study measured middle cerebral artery blood flow lateralization using Functional Transcranial Doppler (fTCD) ultrasonography during the performance of arithmetic tasks to test the triple-code model of numerical functioning. It was hypothesised that blood flow would be greater to the hemisphere responsible for language during multiplication tasks which depend on the retrieval of stored semantic knowledge, whereas bilateral activation would be more likely when adults performed subtraction. Thirteen adults, aged between 18 and 37 years, participated in a one-day cognitive and neuroimaging research program hosted at the Neurocognitive Development Unit. Laterality Indices were obtained whilst they performed an expressive language task and both multiplication and subtraction tasks. Blood flow during the expressive language and numerical calculation tasks showed statistically significant leftward lateralizations and the lateralization indices obtained during the mathematics tasks did not differ significantly from that obtained during the expressive language task (Cohen's d for repeated measures d = 0.160 and d = 0.290 for the multiplication and subtraction tasks, respectively). In partial support of the triple-code model, blood flow while performing multiplication operations was largest over the hemisphere corresponding to language generation. However, contrary to our expectations, blood flow was also significantly lateralised to the left hemisphere while participants performed subtraction. One plausible explanation for these results is that simple subtraction operations similarly draw on the semantic store. It is also possible that the response requirement for both tasks was a confounding factor and future research will attempt to control for this potential confound. Keywords: Arithmetic, subtraction, multiplication, fTCD, middle cerebral artery (MCA), Triple-Code Model Conference: XII International Conference on Cognitive Neuroscience (ICON-XII), Brisbane, Queensland, Australia, 27 Jul - 31 Jul, 2014. Presentation Type: Poster Topic: Cognition and Executive Processes Citation: Connaughton V, Bothma V, Amiruddin A, Clunies-Ross K, French N and Fox A (2015). Hemispheric Specialization for Processing Arithmetic in Adults. Conference Abstract: XII International Conference on Cognitive Neuroscience (ICON-XII). doi: 10.3389/conf.fnhum.2015.217.00128 Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters. The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated. Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed. For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions. Received: 19 Feb 2015; Published Online: 24 Apr 2015. * Correspondence: Ms. Veronica Connaughton, Univeristy of Western Australia, School of Psychology, Perth, Australia, 21104481@STUDENT.UWA.EDU.AU Login Required This action requires you to be registered with Frontiers and logged in. To register or login click here. Abstract Info Abstract The Authors in Frontiers Veronica Connaughton Vicole Bothma Azhani Amiruddin Karen Clunies-Ross Noel French Allison Fox Google Veronica Connaughton Vicole Bothma Azhani Amiruddin Karen Clunies-Ross Noel French Allison Fox Google Scholar Veronica Connaughton Vicole Bothma Azhani Amiruddin Karen Clunies-Ross Noel French Allison Fox PubMed Veronica Connaughton Vicole Bothma Azhani Amiruddin Karen Clunies-Ross Noel French Allison Fox Related Article in Frontiers Google Scholar PubMed Abstract Close Back to top Javascript is disabled. Please enable Javascript in your browser settings in order to see all the content on this page.