arXiv subjects

Yu Zhang

Publications and source records attributed to Yu Zhang.

At least 1,063 records · Page 59Linked to original sources

arXiv2023

Effective Seed-Guided Topic Discovery by Integrating Multiple Types of Contexts

Instead of mining coherent topics from a given text corpus in a completely unsupervised manner, seed-guided topic discovery methods leverage user-provided seed words to extract distinctive and coherent topics so that the mined topics can better cater to the user's interest. To model the semantic correlation between words and seeds for discovering topic-indicative terms, existing seed-guided approaches utilize different types of context signals, such as document-level word co-occurrences, sliding window-based local contexts, and generic linguistic knowledge brought by pre-trained language models. In this work, we analyze and show empirically that each type of context information has its value and limitation in modeling word semantics under seed guidance, but combining three types of contexts (i.e., word embeddings learned from local contexts, pre-trained language model representations obtained from general-domain training, and topic-indicative sentences retrieved based on seed information) allows them to complement each other for discovering quality topics. We propose an iterative framework, SeedTopicMine, which jointly learns from the three types of contexts and gradually fuses their context signals via an ensemble ranking process. Under various sets of seeds and on multiple datasets, SeedTopicMine consistently yields more coherent and accurate topics than existing seed-guided topic discovery approaches.

cs.CL↗

arXiv2023

Deep Learning of Near Field Beam Focusing in Terahertz Wideband Massive MIMO Systems

Employing large antenna arrays and utilizing large bandwidth have the potential of bringing very high data rates to future wireless communication systems. However, this brings the system into the near-field regime and also makes the conventional transceiver architectures suffer from the wideband effects. To address these problems, in this paper, we propose a low-complexity frequency-aware beamforming solution that is designed for hybrid time-delay and phase-shifter based RF architectures. To reduce the complexity, the joint design problem of the time delays and phase shifts is decomposed into two subproblems, where a signal model inspired online learning framework is proposed to learn the shifts of the quantized analog phase shifters, and a low-complexity geometry-assisted method is leveraged to configure the delay settings of the time-delay units. Simulation results highlight the efficacy of the proposed solution in achieving robust performance across a wide frequency range for large antenna array systems.

cs.IT↗

arXiv2022

How swimming style affects schooling of two fish-like wavy hydrofoils

Fish swimming style and schooling behaviour are two important aspects of underwater locomotion. How swimming style affects fish schooling is investigated by a representative problem setup consisting of two tethered NACA0012 hydrofoils undulating in a free stream flow at various wavelengths $ λ^* = 0.8 - 8 $, front-back distance $ D = 0,\ 0.25,\ 0.5,\ 0.75 $, phase difference $ ϕ/π= 0,\ 0.5,\ 1,\ 1.5 $, and lateral gap distance $ G = 0.25,\ 0.3,\ 0.35 $ with fixed Reynolds number $ Re = 5000 $, Strouhal number $ St = 0.4 $, and maximum amplitude $ A_{max} = 0.1 $. In total, 336 cases were simulated by open-source software IBAMR based on immersed boundary method. The increase in wavelength generally leads to higher thrust and lateral force, consistent with a previous study regarding a single swimmer. The highest Froude efficiency is obtained at intermediate wavelength $ λ^* = 2 $ for either individual or two foils as a group. At side-by-side arrangement $ D = 0 $, the thrust force upon the two foils can be equivalent, indicating a stable formation. At staggered arrangement, $ D > 0 $, the follower can take significant advantage of the leader in locomotion performance by tuning phase difference. The follower's benefit decreases with relative distance. Various combinations of wavelength and relative distance can lead to distinct flow structures, indicating a tunable stealth capacity of the schooling swimmers.

physics.flu-dyn↗

arXiv2022

Finite temperature QCD phase transition with 3 flavors of Möbius domain wall fermions

We investigate the finite temperature QCD phase transition with three degenerate quark flavors using Möbius domain wall fermions. To explore the order of phase transition on the lower left corner of Columbia plot and if possible, to locate the critical endpoint we performed simulations at temperatures around 181 and 121 MeV with lattice spacing $a=0.1361(20)$ fm corresponding to temporal lattice extent $N_τ=8,12$ with varying quark mass for two different volumes with aspect ratios $N_σ/N_τ$ ranging from 2 to 3. By analyzing the volume and mass dependence of the chiral condensate, disconnected chiral susceptibility and Binder cumulant we find that there is a crossover at $m_q^{\mathrm{\overline {MS}}}(2\, \mathrm{GeV}) \sim 44\, \mathrm{MeV}$ for $\mathrm{T_{pc}}\sim$ 181 MeV, At temperature 121 MeV, the binder cumulant suggests a crossover at $m_q^{\mathrm{\overline {MS}}}(2\, \mathrm{GeV}) \sim 3.7\, \mathrm{MeV}$, although a study of volume dependence would be important to confirm this.

hep-lat↗

arXiv2022

Enabling Temporal-Spectral Decoding in Pre-movement Detection

Non-invasive brain-computer interfaces help the subjects to control external devices by brain intentions. The multi-class classification of upper limb movements can provide external devices with more control commands. The onsets of the upper limb movements are located by the external limb trajectory to eliminate the delay and bias among the trials. However, the trajectories are not recorded due to the limitation of experiments. The delay cannot be avoided in the analysis of signals. The delay negatively influences the classification performance, which limits the further application of upper limb movements in the brain-computer interface. This work focuses on multi-channel brain signals analysis in the temporal-frequency approach. It proposes the two-stage-training temporal-spectral neural network (TTSNet) to decode patterns from brain signals. The TTSNet first divides the signals into various filter banks. In each filter bank, task-related component analysis is used to reduce the dimension and reject the noise of the brain. A convolutional neural network (CNN) is then used to optimize the temporal characteristic of signals and extract class-related features. Finally, these class-related features from all filter banks are fused by concatenation and classified by the fully connected layer of the CNN. The proposed method is evaluated in two public datasets. The results show that TTSNet has an improved accuracy of 0.7456$\pm$0.1205 compared to the EEGNet of 0.6506$\pm$0.1275 ($p<0.05$) and FBTRCA of 0.6787$\pm$0.1260 ($p<0.1$) in the movement detection task, which classifies the movement state and the resting state. The proposed method is expected to help detect limb movements and assist in the rehabilitation of stroke patients.

cs.HC↗

arXiv2022

Dynamic Sparse Network for Time Series Classification: Learning What to "see''

The receptive field (RF), which determines the region of time series to be ``seen'' and used, is critical to improve the performance for time series classification (TSC). However, the variation of signal scales across and within time series data, makes it challenging to decide on proper RF sizes for TSC. In this paper, we propose a dynamic sparse network (DSN) with sparse connections for TSC, which can learn to cover various RF without cumbersome hyper-parameters tuning. The kernels in each sparse layer are sparse and can be explored under the constraint regions by dynamic sparse training, which makes it possible to reduce the resource cost. The experimental results show that the proposed DSN model can achieve state-of-art performance on both univariate and multivariate TSC datasets with less than 50\% computational cost compared with recent baseline methods, opening the path towards more accurate resource-aware methods for time series analyses. Our code is publicly available at: https://github.com/QiaoXiao7282/DSN.

cs.LG↗

arXiv2022

Universal Paralinguistic Speech Representations Using Self-Supervised Conformers

Many speech applications require understanding aspects beyond the words being spoken, such as recognizing emotion, detecting whether the speaker is wearing a mask, or distinguishing real from synthetic speech. In this work, we introduce a new state-of-the-art paralinguistic representation derived from large-scale, fully self-supervised training of a 600M+ parameter Conformer-based architecture. We benchmark on a diverse set of speech tasks and demonstrate that simple linear classifiers trained on top of our time-averaged representation outperform nearly all previous results, in some cases by large margins. Our analyses of context-window size demonstrate that, surprisingly, 2 second context-windows achieve 96\% the performance of the Conformers that use the full long-term context on 7 out of 9 tasks. Furthermore, while the best per-task representations are extracted internally in the network, stable performance across several layers allows a single universal representation to reach near optimal performance on all tasks.

cs.SD↗

arXiv2022

Unsupervised Deep Learning for AC Optimal Power Flow via Lagrangian Duality

Non-convex AC optimal power flow (AC-OPF) is a fundamental optimization problem in power system analysis. The computational complexity of conventional solvers is typically high and not suitable for large-scale networks in real-time operation. Hence, deep learning based approaches have gained intensive attention to conduct the time-consuming training process offline. Supervised learning methods may yield a feasible AC-OPF solution with a small optimality gap. However, they often need conventional solvers to generate the training dataset. This paper proposes an end-to-end unsupervised learning based framework for AC-OPF. We develop a deep neural network to output a partial set of decision variables while the remaining variables are recovered by solving AC power flow equations. The fast decoupled power flow solver is adopted to further reduce the computational time. In addition, we propose using a modified augmented Lagrangian function as the training loss. The multipliers are adjusted dynamically based on the degree of constraint violation. Extensive numerical test results corroborate the advantages of our proposed approach over some existing methods.

eess.SY↗

arXiv2022

Entity Set Co-Expansion in StackOverflow

Given a few seed entities of a certain type (e.g., Software or Programming Language), entity set expansion aims to discover an extensive set of entities that share the same type as the seeds. Entity set expansion in software-related domains such as StackOverflow can benefit many downstream tasks (e.g., software knowledge graph construction) and facilitate better IT operations and service management. Meanwhile, existing approaches are less concerned with two problems: (1) How to deal with multiple types of seed entities simultaneously? (2) How to leverage the power of pre-trained language models (PLMs)? Being aware of these two problems, in this paper, we study the entity set co-expansion task in StackOverflow, which extracts Library, OS, Application, and Language entities from StackOverflow question-answer threads. During the co-expansion process, we use PLMs to derive embeddings of candidate entities for calculating similarities between entities. Experimental results show that our proposed SECoExpan framework outperforms previous approaches significantly.

cs.CL↗

arXiv2022

Feature Aggregation and Propagation Network for Camouflaged Object Detection

Camouflaged object detection (COD) aims to detect/segment camouflaged objects embedded in the environment, which has attracted increasing attention over the past decades. Although several COD methods have been developed, they still suffer from unsatisfactory performance due to the intrinsic similarities between the foreground objects and background surroundings. In this paper, we propose a novel Feature Aggregation and Propagation Network (FAP-Net) for camouflaged object detection. Specifically, we propose a Boundary Guidance Module (BGM) to explicitly model the boundary characteristic, which can provide boundary-enhanced features to boost the COD performance. To capture the scale variations of the camouflaged objects, we propose a Multi-scale Feature Aggregation Module (MFAM) to characterize the multi-scale information from each layer and obtain the aggregated feature representations. Furthermore, we propose a Cross-level Fusion and Propagation Module (CFPM). In the CFPM, the feature fusion part can effectively integrate the features from adjacent layers to exploit the cross-level correlations, and the feature propagation part can transmit valuable context information from the encoder to the decoder network via a gate unit. Finally, we formulate a unified and end-to-end trainable framework where cross-level features can be effectively fused and propagated for capturing rich context information. Extensive experiments on three benchmark camouflaged datasets demonstrate that our FAP-Net outperforms other state-of-the-art COD models. Moreover, our model can be extended to the polyp segmentation task, and the comparison results further validate the effectiveness of the proposed model in segmenting polyps. The source code and results will be released at https://github.com/taozh2017/FAPNet.

cs.CV↗

arXiv2022

ExpNet: A unified network for Expert-Level Classification

Different from the general visual classification, some classification tasks are more challenging as they need the professional categories of the images. In the paper, we call them expert-level classification. Previous fine-grained vision classification (FGVC) has made many efforts on some of its specific sub-tasks. However, they are difficult to expand to the general cases which rely on the comprehensive analysis of part-global correlation and the hierarchical features interaction. In this paper, we propose Expert Network (ExpNet) to address the unique challenges of expert-level classification through a unified network. In ExpNet, we hierarchically decouple the part and context features and individually process them using a novel attentive mechanism, called Gaze-Shift. In each stage, Gaze-Shift produces a focal-part feature for the subsequent abstraction and memorizes a context-related embedding. Then we fuse the final focal embedding with all memorized context-related embedding to make the prediction. Such an architecture realizes the dual-track processing of partial and global information and hierarchical feature interactions. We conduct the experiments over three representative expert-level classification tasks: FGVC, disease classification, and artwork attributes classification. In these experiments, superior performance of our ExpNet is observed comparing to the state-of-the-arts in a wide range of fields, indicating the effectiveness and generalization of our ExpNet. The code will be made publicly available.

cs.CV↗

arXiv2022

TSGP: Two-Stage Generative Prompting for Unsupervised Commonsense Question Answering

Unsupervised commonsense question answering requires mining effective commonsense knowledge without the rely on the labeled task data. Previous methods typically retrieved from traditional knowledge bases or used pre-trained language models (PrLMs) to generate fixed types of knowledge, which have poor generalization ability. In this paper, we aim to address the above limitation by leveraging the implicit knowledge stored in PrLMs and propose a two-stage prompt-based unsupervised commonsense question answering framework (TSGP). Specifically, we first use knowledge generation prompts to generate the knowledge required for questions with unlimited types and possible candidate answers independent of specified choices. Then, we further utilize answer generation prompts to generate possible candidate answers independent of specified choices. Experimental results and analysis on three different commonsense reasoning tasks, CommonsenseQA, OpenBookQA, and SocialIQA, demonstrate that TSGP significantly improves the reasoning ability of language models in unsupervised settings. Our code is available at: https://github.com/Yueqing-Sun/TSGP.

cs.CL↗

arXiv2022

Boosting Thermoelectric Power Factor of Carbon Nanotube Networks with Excluded Volume by Co-embedded Microparticles

Carbon nanotube (CNTs) networks embedded in polymer matrix have been extensively studied over the recent years as a flexible thermoelectric (TE) transport medium. However, their power factor has been largely limited by the relatively inefficient tunneling transport at junctions between CNTs and the low density of conducting channels through the networks. In this work, we demonstrate that significant enhancement of power factor is possible by adding electrically insulating microscale particles in CNT networks. When silica particles of a few micrometer diameters were co-embedded in single-walled CNT-polydimethylsiloxane (PDMS) composites, both the electrical conductivity and the Seebeck coefficient were simultaneously enhanced, thereby boosting the power factor by more than a factor of six. We find that the silica microparticles excluded a large volume of the composite from the access of CNTs and caused CNT networks to form around them using the polymer as a binder, which in turn resulted in improved connectivity and alignment of CNTs. Our theoretical calculations based on junction tunneling transport show that the large power factor enhancement can be attributed to the enhanced tunneling with reduced junction distance between CNTs and the increased geometric factor due to better CNT alignment. Additional enhancement of power factor by more than a factor of two was achieved by sample compression due to the further improvement of CNT alignment.

physics.app-ph↗

arXiv2022

TLP: A Deep Learning-based Cost Model for Tensor Program Tuning

Tensor program tuning is a non-convex objective optimization problem, to which search-based approaches have proven to be effective. At the core of the search-based approaches lies the design of the cost model. Though deep learning-based cost models perform significantly better than other methods, they still fall short and suffer from the following problems. First, their feature extraction heavily relies on expert-level domain knowledge in hardware architectures. Even so, the extracted features are often unsatisfactory and require separate considerations for CPUs and GPUs. Second, a cost model trained on one hardware platform usually performs poorly on another, a problem we call cross-hardware unavailability. In order to address these problems, we propose TLP and MTLTLP. TLP is a deep learning-based cost model that facilitates tensor program tuning. Instead of extracting features from the tensor program itself, TLP extracts features from the schedule primitives. We treat schedule primitives as tensor languages. TLP is thus a Tensor Language Processing task. In this way, the task of predicting the tensor program latency through the cost model is transformed into a natural language processing (NLP) regression task. MTL-TLP combines Multi-Task Learning and TLP to cope with the cross-hardware unavailability problem. We incorporate these techniques into the Ansor framework and conduct detailed experiments. Results show that TLP can speed up the average search time by 9.1X and 3.0X on CPU and GPU workloads, respectively, compared to the state-of-the-art implementation. MTL-TLP can achieve a speed-up of 4.7X and 2.9X on CPU and GPU workloads, respectively, using only 7% of the target hardware data.

cs.LG↗

arXiv2022

Molecular Dynamics Study of Plasmon-Mediated Chemical Transformations

Heterogeneous catalysis of adsorbates on metallic surfaces mediated by plasmon has potential high photoelectric conversion efficiency and controllable reaction selectivity. Theoretical modeling of dynamical reaction processes provides in-depth analyses complementing experimental investigations. Especially for plasmon-mediated chemical transformations, light absorption, photoelectric conversion, electron-electron scattering, and electron-phonon coupling occur simultaneously at different timescales, rendering it very challenging to delineate the complex interplay of different factors. In this work, a trajectory surface hopping non-adiabatic molecular dynamics method is used to investigate the dynamics of plasmon excitation in an Au$_{20}$-CO system, including hot carrier generation, plasmon energy relaxation, and CO activation induced by electron-vibration coupling. The electronic properties indicate that when Au$_{20}$-CO is excited, a partial charge transfer takes place from Au$_{20}$ to CO. On the other hand, the dynamical simulations show that hot carriers generated after plasmon excitation transfer back and forth between Au$_{20}$ and CO. Meanwhile, the C-O stretching mode is activated due to the non-adiabatic couplings. The efficiency of plasmon-mediated transformation ($\sim$40\%) is obtained based on the ensemble average of these quantities. Our simulations provide important dynamical and atomistic insights into plasmon-mediated chemical transformation from the perspective of non-adiabatic simulations.

physics.chem-ph↗

arXiv2022

Disentangling Task Relations for Few-shot Text Classification via Self-Supervised Hierarchical Task Clustering

Few-Shot Text Classification (FSTC) imitates humans to learn a new text classifier efficiently with only few examples, by leveraging prior knowledge from historical tasks. However, most prior works assume that all the tasks are sampled from a single data source, which cannot adapt to real-world scenarios where tasks are heterogeneous and lie in different distributions. As such, existing methods may suffer from their globally knowledge-shared mechanisms to handle the task heterogeneity. On the other hand, inherent task relation are not explicitly captured, making task knowledge unorganized and hard to transfer to new tasks. Thus, we explore a new FSTC setting where tasks can come from a diverse range of data sources. To address the task heterogeneity, we propose a self-supervised hierarchical task clustering (SS-HTC) method. SS-HTC not only customizes cluster-specific knowledge by dynamically organizing heterogeneous tasks into different clusters in hierarchical levels but also disentangles underlying relations between tasks to improve the interpretability. Extensive experiments on five public FSTC benchmark datasets demonstrate the effectiveness of SS-HTC.

cs.CL↗

arXiv2022

Normative Modeling via Conditional Variational Autoencoder and Adversarial Learning to Identify Brain Dysfunction in Alzheimer's Disease

Normative modeling is an emerging and promising approach to effectively study disorder heterogeneity in individual participants. In this study, we propose a novel normative modeling method by combining conditional variational autoencoder with adversarial learning (ACVAE) to identify brain dysfunction in Alzheimer's Disease (AD). Specifically, we first train a conditional VAE on the healthy control (HC) group to create a normative model conditioned on covariates like age, gender and intracranial volume. Then we incorporate an adversarial training process to construct a discriminative feature space that can better generalize to unseen data. Finally, we compute deviations from the normal criterion at the patient level to determine which brain regions were associated with AD. Our experiments on OASIS-3 database show that the deviation maps generated by our model exhibit higher sensitivity to AD compared to other deep normative models, and are able to better identify differences between the AD and HC groups.

cs.LG↗

arXiv2022

Tensor-based Intrinsic Subspace Representation Learning for Multi-view Clustering

As a hot research topic, many multi-view clustering approaches are proposed over the past few years. Nevertheless, most existing algorithms merely take the consensus information among different views into consideration for clustering. Actually, it may hinder the multi-view clustering performance in real-life applications, since different views usually contain diverse statistic properties. To address this problem, we propose a novel Tensor-based Intrinsic Subspace Representation Learning (TISRL) for multi-view clustering in this paper. Concretely, the rank preserving decomposition is proposed firstly to effectively deal with the diverse statistic information contained in different views. Then, to achieve the intrinsic subspace representation, the tensor-singular value decomposition based low-rank tensor constraint is also utilized in our method. It can be seen that specific information contained in different views is fully investigated by the rank preserving decomposition, and the high-order correlations of multi-view data are also mined by the low-rank tensor constraint. The objective function can be optimized by an augmented Lagrangian multiplier based alternating direction minimization algorithm. Experimental results on nine common used real-world multi-view datasets illustrate the superiority of TISRL.

cs.LG↗