Molecular dynamics (MD) simulation techniques are widely used for various natural science applications. Increasingly, machine learning (ML) force field (FF) models begin to replace ab-initio simulations by predicting forces directly from atomic structures. Despite significant progress in this area, such primarily benchmarked their force/energy prediction errors, even though the practical use case would be produce realistic MD trajectories. We aim fill gap introducing a novel benchmark suite...

In this study we deal with the problem of identifying and categorizing offensive language in social media. Our group, BNU-HKBU UIC NLP Team2, use supervised classification along multiple version data generated by different ways pre-processing data. We then state-of-the-art model Bidirectional Encoder Representations from Transformers, or BERT (Devlin et al, 2018), to capture linguistic, syntactic semantic features. Long range dependencies between each part a sentence can be captured BERT’s...

The construction of health indicators (HI) for traditional deep learning requires human training labels and poor interpretability. This paper proposes an HI method based on Stacked Sparse Autoencoder (SSAE) combines SSAE with Long short-term memory (LSTM) network to predict the remaining useful life (RUL). Extracting features from a single domain may result in insufficient feature extraction cannot comprehensively reflect degradation status information mechanical equipment. In order solve...

The increase in the complexity of modern electronic products has brought significant challenges to fault diagnosis circuits, and current methods have problems such as long identification time, inaccurate positioning, low diagnostic efficiency. In response these situations. This paper proposes a method for circuits combining synchronous synchrosqueezing wavelet transform (SWT), deep convolutional neural network (DCNN), extreme learning machine (ELM). First, original signal is noise-reduced...

The effect of NH3 plasma treatment on the interfacial property between ultrathin HfO2 and strained Si0.65Ge0.35 substrate has been investigated by high-resolution cross-sectional transmission electron micrographs, x-ray photoelectron spectroscopy, VBS, capacitance-voltage (C-V), current density-voltage (J-V). TEM XPS results confirm that layer with N–Hf N–Si/O-N-Si bonds acts as a barrier against interdiffusions during annealing in some degree. valence-band offsets ΔEV HfO2/SiGe interfaces...

In a complex working environment, the fault signal of gearbox is greatly affected by outside world, and feature recognition difficult, so diagnosis accuracy difficult to meet expected requirements. To solve this problem, paper proposes method based on an optimized stacked denoising auto encoder (SDAE) kernel extreme learning machine (KELM). Firstly, Particle Swarm Optimization algorithm in Adaptive Weight (SAPSO) was adopted optimize SDAE network structure, number hidden layer nodes, rate,...

Fixed length summarization aims at generating summaries with a preset number of words or characters.Most recent researches incorporate information word embeddings as the input to recurrent decoding unit, causing compromise between controllability and summary quality.In this work, we present an effective controlling unit Length Attention (LenAtten) break trade-off.Experimental results show that LenAtten not only brings improvements in ROGUE scores but also has great generalization ability.In...

Aim: Rapid economic development and urban expansion have caused dramatic changes in land use Guangdong-Hong Kong-Macao Greater Bay Area (GHMGBA), resulting the urgent need to study their effects on distributions habitat connectivity of birds this area.Method: Based 15 environmental variables, we used maximum entropy (MaxEnt) model predict habitats 82 bird species GHMGBA 2000 2015, then analyzed distribution patterns under urbanization using an equivalent connected area index (ECA).Results:...

Abstract To solve the problem of feature extraction in electronic circuits due to nonstationary and nonlinear characteristics fault signals, a method for is proposed, which combines wavelet packet analysis an improved landmark ISOMAP mapping algorithm. The technology used decompose reconstruct signals at multiple levels. extracted entropy construct vector matrix. density-based spatial clustering applications with noise (DBSCAN) algorithm calculate screen points. embed high-dimensional...

Missing values are a common problem that poses significant challenges to data analysis and machine learning. This necessitates the development of an effective imputation method fill in missing accurately, thereby enhancing overall quality utility datasets. Existing methods, however, fall short explicitly considering `missingness' information during embedding initialization stage modeling entangled feature sample correlations learning process, thus leading inferior performance. We propose...

Semi-supervised learning (SSL) has been widely used to learn from both a few labeled images and many unlabeled overcome the scarcity of samples in medical image segmentation. Most current SSL-based segmentation methods use pixel values directly identify similar features data. They usually fail accurately capture intricate attachment structures left atrium, such as areas inconsistent density or exhibit outward curvatures, adding complexity task. In this paper, we delve into issue introduce an...

In this study, we take two batches of 323 provincial traditional villages in Shaanxi Province from 2015 to 2017 as the research object. We use Google Earth conduct spatial positioning, analyze villages’ distribution characteristics, and carry out correlation analysis under different geographical units influencing factors. The study shows that have three high-density core areas, with increase altitude terrain undulation, its a "low–high–low" pattern; mostly concentrated...

The effect of NH3 plasma treatment on the interfacial and dielectric property between ultrathin HfLaO strained Si0.65Ge0.35 substrate is investigated by high-resolution cross-sectional transmission electron micrographs (HR-TEM), Auger spectroscopy (AES), x-ray photoelectron (XPS), near edge absorption fine structure (NEXAFS), valence-band spectra current density–voltage (J–V). TEM, AES XPS results confirm that layer with N–Hf, La–N Si–N/Si–O–N bonds acts as a barrier against interdiffusions...

The insights gained from individual tracking can improve the conservation of released young green sea turtles by providing detailed data on migration route, feeding place and habitat distribution. During fishing moratorium May 1 to August 16 in 2021, we used satellites track 6 juvenile China’s Huidong Sea Turtle Nature Reserve, located reconstructed their routes foraging areas. shows that longest time is 91 days, shortest 10 total 213 which three (50%) obtained for more than one month,...

Abstract With the rapid development of science and technology, computer technology has been applied to all walks life, including English translation. Through translation, we can greatly improve efficiency translation production, which an unprecedented impact on language service industry. At same time, with global economy, economy society put forward higher requirements for industry, requires us constantly quality speed Therefore, in order better brought a significant basis it. However, due...

Learning pair interactions from experimental or simulation data is of great interest for molecular simulations. We propose a general stochastic method learning using differentiable simulations (DiffSim). DiffSim defines loss function based on structural observables, such as the radial distribution function, through dynamics (MD) The interaction potentials are then learned directly by gradient descent, backpropagation to calculate metric with respect potential MD simulation. This...

Matrix-free nanocomposites made from polymer grafted nanoparticles (PGN) represent a paradigm shift in materials science because they greatly improve nanoparticle dispersion and offer greater tunability over rheological mechanical properties comparison to neat polymers. Utilizing the full potential of PGNs requires deeper understanding how graft length, density, chemistry influence interfacial interactions between particles. There has been great progress describing these effects with...

Fixed length summarization aims at generating summaries with a preset number of words or characters. Most recent researches incorporate information word embeddings as the input to recurrent decoding unit, causing compromise between controllability and summary quality. In this work, we present an effective controlling unit Length Attention (LenAtten) break trade-off. Experimental results show that LenAtten not only brings improvements in ROGUE scores but also has great generalization ability....