arXiv subjects

Peng Tao

Publications and source records attributed to Peng Tao.

16 recordsLinked to original sources

arXiv2025

AWEMixer: Adaptive Wavelet-Enhanced Mixer Network for Long-Term Time Series Forecasting

Forecasting long-term time series in IoT environments remains a significant challenge due to the non-stationary and multi-scale characteristics of sensor signals. Furthermore, error accumulation causes a decrease in forecast quality when predicting further into the future. Traditional methods are restricted to operate in time-domain, while the global frequency information achieved by Fourier transform would be regarded as stationary signals leading to blur the temporal patterns of transient events. We propose AWEMixer, an Adaptive Wavelet-Enhanced Mixer Network including two innovative components: 1) a Frequency Router designs to utilize the global periodicity pattern achieved by Fast Fourier Transform to adaptively weight localized wavelet subband, and 2) a Coherent Gated Fusion Block to achieve selective integration of prominent frequency features with multi-scale temporal representation through cross-attention and gating mechanism, which realizes accurate time-frequency localization while remaining robust to noise. Seven public benchmarks validate that our model is more effective than recent state-of-the-art models. Specifically, our model consistently achieves performance improvement compared with transformer-based and MLP-based state-of-the-art models in long-sequence time series forecasting. Code is available at https://github.com/hit636/AWEMixer

cs.LG↗

arXiv2025

A Time-Series Foundation Model by Universal Delay Embedding

This study introduces Universal Delay Embedding (UDE), a pretrained foundation model designed to revolutionize time-series forecasting through principled integration of delay embedding representation and Koopman operator prediction. Leveraging Takens' embedding theorem, UDE as a dynamical representation of observed data constructs two-dimensional subspace patches from Hankel matrices, theoretically preserving dynamical and topological properties of underlying dynamical systems. Such patches are viewed as images, which can be efficiently processed by exploiting advanced deep learning technologies. Computationally, these patches further serve as tokens for learning a self-attention encoder, thus enabling accurate prediction of nonlinear time-series by a finite-dimensional Koopman operator in a linear manner in a latent space. Extensive evaluations across various benchmarks and real-world climate datasets demonstrate over 20% average reduction in mean squared error versus state-of-the-art foundation models, alongside superior generalization in fine-tuning scenarios. In particular, the learned dynamical representations and Koopman operator prediction forms from the patches exhibit exceptional interpretability, with consistent identification of topologically informative subspaces and robust encoding of domain-invariant dynamics, establishing UDE as a scalable, interpretable framework for universal time-series modeling and forecasting with broad scientific and industrial applicability.

cs.LG↗

arXiv2025

Delayformer: spatiotemporal transformation for predicting high-dimensional dynamics

Predicting time-series is of great importance in various scientific and engineering fields. However, in the context of limited and noisy data, accurately predicting dynamics of all variables in a high-dimensional system is a challenging task due to their nonlinearity and also complex interactions. Current methods including deep learning approaches often perform poorly for real-world systems under such circumstances. This study introduces the Delayformer framework for simultaneously predicting dynamics of all variables, by developing a novel multivariate spatiotemporal information (mvSTI) transformation that makes each observed variable into a delay-embedded state (vector) and further cross-learns those states from different variables. From dynamical systems viewpoint, Delayformer predicts system states rather than individual variables, thus theoretically and computationally overcoming such nonlinearity and cross-interaction problems. Specifically, it first utilizes a single shared Visual Transformer (ViT) encoder to cross-represent dynamical states from observed variables in a delay embedded form and then employs distinct linear decoders for predicting next states, i.e. equivalently predicting all original variables parallelly. By leveraging the theoretical foundations of delay embedding theory and the representational capabilities of Transformers, Delayformer outperforms current state-of-the-art methods in forecasting tasks on both synthetic and real-world datasets. Furthermore, the potential of Delayformer as a foundational time-series model is demonstrated through cross-domain forecasting tasks, highlighting its broad applicability across various scenarios.

cs.LG↗

arXiv2024

Brain-inspired Chaotic Graph Backpropagation for Large-scale Combinatorial Optimization

Graph neural networks (GNNs) with unsupervised learning can solve large-scale combinatorial optimization problems (COPs) with efficient time complexity, making them versatile for various applications. However, since this method maps the combinatorial optimization problem to the training process of a graph neural network, and the current mainstream backpropagation-based training algorithms are prone to fall into local minima, the optimization performance is still inferior to the current state-of-the-art (SOTA) COP methods. To address this issue, inspired by possibly chaotic dynamics of real brain learning, we introduce a chaotic training algorithm, i.e. chaotic graph backpropagation (CGBP), which introduces a local loss function in GNN that makes the training process not only chaotic but also highly efficient. Different from existing methods, we show that the global ergodicity and pseudo-randomness of such chaotic dynamics enable CGBP to learn each optimal GNN effectively and globally, thus solving the COP efficiently. We have applied CGBP to solve various COPs, such as the maximum independent set, maximum cut, and graph coloring. Results on several large-scale benchmark datasets showcase that CGBP can outperform not only existing GNN algorithms but also SOTA methods. In addition to solving large-scale COPs, CGBP as a universal learning algorithm for GNNs, i.e. as a plug-in unit, can be easily integrated into any existing method for improving the performance.

cs.LG↗

arXiv2023

PASSerRank: Prediction of Allosteric Sites with Learning to Rank

Allostery plays a crucial role in regulating protein activity, making it a highly sought-after target in drug development. One of the major challenges in allosteric drug research is the identification of allosteric sites. In recent years, many computational models have been developed for accurate allosteric site prediction. Most of these models focus on designing a general rule that can be applied to pockets of proteins from various families. In this study, we present a new approach using the concept of Learning to Rank (LTR). The LTR model ranks pockets based on their relevance to allosteric sites, i.e., how well a pocket meets the characteristics of known allosteric sites. The model outperforms other common machine learning models with higher F1 score and Matthews correlation coefficient. After the training and validation on two datasets, the Allosteric Database (ASD) and CASBench, the LTR model was able to rank an allosteric pocket in the top 3 positions for 83.6% and 80.5% of test proteins, respectively. The trained model is available on the PASSer platform (https://passer.smu.edu) to aid in drug discovery research.

q-bio.QM↗

arXiv2022

Accurate ADMET Prediction with XGBoost

The absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties are important in drug discovery as they define efficacy and safety. In this work, we applied an ensemble of features, including fingerprints and descriptors, and a tree-based machine learning model, extreme gradient boosting, for accurate ADMET prediction. Our model performs well in the Therapeutics Data Commons ADMET benchmark group. For 22 tasks, our model is ranked first in 18 tasks and top 3 in 21 tasks. The trained machine learning models are integrated in ADMETboost, a web server that is publicly available at https://ai-druglab.smu.edu/admet.

q-bio.BM↗

arXiv2022

LAST: Latent Space Assisted Adaptive Sampling for Protein Trajectories

Molecular dynamics (MD) simulation is widely used to study protein conformations and dynamics. However, conventional simulation suffers from being trapped in some local energy minima that are hard to escape. Thus, most computational time is spent sampling in the already visited regions. This leads to an inefficient sampling process and further hinders the exploration of protein movements in affordable simulation time. The advancement of deep learning provides new opportunities for protein sampling. Variational autoencoders are a class of deep learning models to learn a low-dimensional representation (referred to as the latent space) that can capture the key features of the input data. Based on this characteristic, we proposed a new adaptive sampling method, latent space assisted adaptive sampling for protein trajectories (LAST), to accelerate the exploration of protein conformational space. This method comprises cycles of (i) variational autoencoders training, (ii) seed structure selection on the latent space and (iii) conformational sampling through additional MD simulations. The proposed approach is validated through the sampling of four structures of two protein systems: two metastable states of E. Coli adenosine kinase (ADK) and two native states of Vivid (VVD). In all four conformations, seed structures were shown to lie on the boundary of conformation distributions. Moreover, large conformational changes were observed in a shorter simulation time when compared with conventional MD (cMD) simulations in both systems. In metastable ADK simulations, LAST explored two transition paths toward two stable states while cMD became trapped in an energy basin. In VVD light state simulations, LAST was three times faster than cMD simulation with a similar conformational space.

q-bio.QM↗

arXiv2022

Time Series Prediction by Multi-task GPR with Spatiotemporal Information Transformation

Making an accurate prediction of an unknown system only from a short-term time series is difficult due to the lack of sufficient information, especially in a multi-step-ahead manner. However, a high-dimensional short-term time series contains rich dynamical information, and also becomes increasingly available in many fields. In this work, by exploiting spatiotemporal information (STI) transformation scheme that transforms such high-dimensional/spatial information to temporal information, we developed a new method called MT-GPRMachine to achieve accurate prediction from a short-term time series. Specifically, we first construct a specific multi-task GPR which is multiple linked STI mappings to transform high dimensional/spatial information into temporal/dynamical information of any given target variable, and then makes multi step-ahead prediction of the target variable by solving those STI mappings. The multi-step-ahead prediction results on various synthetic and real-world datasets clearly validated that MT-GPRMachine outperformed other existing approaches.

cs.LG↗

arXiv2020

ivis Dimensionality Reduction Framework for Biomacromolecular Simulations

Molecular dynamics (MD) simulations have been widely applied to study macromolecules including proteins. However, high-dimensionality of the datasets produced by simulations makes it difficult for thorough analysis, and further hinders a deeper understanding of biomacromolecules. To gain more insights into the protein structure-function relations, appropriate dimensionality reduction methods are needed to project simulations onto low-dimensional spaces. Linear dimensionality reduction methods, such as principal component analysis (PCA) and time-structure based independent component analysis (t-ICA), could not preserve sufficient structural information. Though better than linear methods, nonlinear methods, such as t-distributed stochastic neighbor embedding (t-SNE), still suffer from the limitations in avoiding system noise and keeping inter-cluster relations. ivis is a novel deep learning-based dimensionality reduction method originally developed for single-cell datasets. Here we applied this framework for the study of light, oxygen and voltage (LOV) domain of diatom Phaeodactylum tricornutum aureochrome 1a (PtAu1a). Compared with other methods, ivis is shown to be superior in constructing Markov state model (MSM), preserving information of both local and global distances and maintaining similarity between high dimension and low dimension with the least information loss. Moreover, ivis framework is capable of providing new prospective for deciphering residue-level protein allostery through the feature weights in the neural network. Overall, ivis is a promising member in the analysis toolbox for proteins.

q-bio.QM↗

arXiv2020

Deciphering the Protein Motion of S1 Subunit in SARS-CoV-2 Spike Glycoprotein Through Integrated Computational Methods

The novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a major worldwide public health emergency that has infected over $1.5$ million people. The partially open state of S1 subunit in spike glycoprotein is considered vital for its infection with host cell and is represented as a key target for neutralizing antibodies. However, the mechanism elucidating the transition from the closed state to the partially open state still remains unclear. Here, we applied a combination of Markov state model, transition path theory and random forest to analyze the S1 motion. Our results explored a promising complete conformational movement of receptor-binding domain, from buried, partially open, to detached states. We also numerically confirmed the transition probability between those states. Based on the asymmetry in both the dynamics behavior and backbone C$α$ importance, we further suggested a relation between chains in the trimer spike protein, which may help in the vaccine design and antibody neutralization.

q-bio.BM↗

arXiv2020

Residual-Recursion Autoencoder for Shape Illustration Images

Shape illustration images (SIIs) are common and important in describing the cross-sections of industrial products. Same as MNIST, the handwritten digit images, SIIs are gray or binary and containing shapes that are surrounded by large areas of blanks. In this work, Residual-Recursion Autoencoder (RRAE) has been proposed to extract low-dimensional features from SIIs while maintaining reconstruction accuracy as high as possible. RRAE will try to reconstruct the original image several times and recursively fill the latest residual image to the reserved channel of the encoder's input before the next trial of reconstruction. As a kind of neural network training framework, RRAE can wrap over other autoencoders and increase their performance. From experiment results, the reconstruction loss is decreased by 86.47% for convolutional autoencoder with high-resolution SIIs, 10.77% for variational autoencoder and 8.06% for conditional variational autoencoder with MNIST.

eess.IV↗

arXiv2017

Spontaneous antiferromagnetic order and strain effect on electronic properties of $α$-graphyne

Using hybrid exchange-correlation functional in ab initio density functional theory calculations, we study magnetic properties and strain effect on the electronic properties of $α$-graphyne monolayer. We find that a spontaneous antiferromagnetic (AF) ordering occurs with energy band gap ($\sim$ 0.5 eV) in the equilibrated $α$-graphyne. Bi-axial tensile strain enhances the stability of AF state as well as the staggered spin moment and value of the energy gap. The antiferromagnetic semiconductor phase is quite robust against moderate carrier filling with threshold carrier density up to 1.7$\times$10$^{14}$ electrons/cm$^2$ to destabilize the phase. The spontaneous AF ordering and strain effect in $α$-graphyne can be well described by the framework of the Hubbard model. Our study shows that it is essential to consider the electronic correlation effect properly in $α$-graphyne and may pave an avenue for exploring magnetic ordering in other carbon allotropes with mixed hybridization of s and p orbitals.

physics.comp-ph↗

arXiv2014

Strain-induced magnetism in MoS2 monolayer with defects

The strain-induced magnetism is observed in single-layer MoS2 with atomic single vacancies from density functional calculations. Calculated magnetic moment is no less than 2muB per vacancy defect. The straininduced band gap closure is concurrent with the occurrence of the magnetism. Possible physical mechanism of the emergence of strain-induced magnetism is illustrated. We also demonstrate the possibility to test the predicted magnetism in experiment. Our study may provide an opportunity for the design of new type of memory-switching or logic devices by using earth-rich nonmagnetic materials MoS2.

cond-mat.mtrl-sci↗

arXiv2013

Skyrmion ground state and gyration of skyrmions in magnetic nanodisks without the Dzyaloshinsky-Moriya interaction

We show by micromagnetic simulations that spontaneous skyrmion ground state can exist in Co/Ru/Co nanodisks without the Dzyaloshinsky-Moriya interaction (DMI), which can remain stable in the applied magnetic field along +z direction even up to 0.44 T. The guiding center ($R_x$,$R_y$) of skyrmion defined by the moments of the topological density presents a novel gyration with a star-like trajectory in a pulsed magnetic field and a hexagonal trajectory after the field is switched off, which is different from that of vortex or bubble. One of the coupled skyrmions could move without an external magnetic field, but only induced by the motion of the other one due to strong inter-layer magnetostatic interactions. This work sheds light on how novel skyrmions can be discovered in various (not limited to magnetic) systems with competing energies and contributes to the understanding of the dynamical properties of skyrmion.

cond-mat.mtrl-sci↗

arXiv2012

High pressure effect on structure, electronic structure and thermoelectric properties of MoS$_2$

We systematically study the effect of high pressure on the structure, electronic structure and transport properties of 2H-MoS$_2$, based on first-principles density functional calculations and the Boltzmann transport theory. Our calculation shows a vanishing anisotropy in the rate of structural change at around 25 GPa, in agreement with the experimental data. A conversion from van der Waals(vdW) to covalent-like bonding is seen. Concurrently, a transition from semiconductor to metal occurs at 25 GPa from band structure calculation. Our transport calculations also find pressure-enhanced electrical conductivities and significant values of the thermoelectric figure of merit over a wide temperature range. Our study supplies a new route to improve the thermoelectric performance of MoS$_2$ and of other transition metal dichalcogenides by applying hydrostatic pressure.

cond-mat.mtrl-sci↗

arXiv2012

Doping effect on thermoelectric properties of MoS$_2$

We systematically study thermoelectric properties of layered MoS$_2$ by doping, based on Boltzmann transport theory and first-principles calculations. We obtain optimal doping region (around 10$^{19}$ cm$^{-3}$) by looking closely to the temperature and doping level dependent thermopower, electrical conductivity, power factor (PF) and ultimately figure of merit (ZT) coefficient along in-plane and cross-plane directions. MoS$_2$ has a vanishingly small anisotropy of thermopower but a big anisotropy of electrical conductivity and electronic thermal conductivity in optimal doping region. $\kappa_e$ is comparable to $\kappa_l$ in the plane while $\kappa_l$ dominates over $\kappa_e$ across the plane. ZT can reach as high as 0.3 at around 700 K. In-plane direction is demonstrated to be more preferable for thermoelectric applications of MoS$_2$ by doping.

cond-mat.mtrl-sci↗