arXiv subjects

Yu Zhang

Publications and source records attributed to Yu Zhang.

At least 1,135 records · Page 63Linked to original sources

arXiv2022

Modern Views of Machine Learning for Precision Psychiatry

In light of the NIMH's Research Domain Criteria (RDoC), the advent of functional neuroimaging, novel technologies and methods provide new opportunities to develop precise and personalized prognosis and diagnosis of mental disorders. Machine learning (ML) and artificial intelligence (AI) technologies are playing an increasingly critical role in the new era of precision psychiatry. Combining ML/AI with neuromodulation technologies can potentially provide explainable solutions in clinical practice and effective therapeutic treatment. Advanced wearable and mobile technologies also call for the new role of ML/AI for digital phenotyping in mobile mental health. In this review, we provide a comprehensive review of the ML methodologies and applications by combining neuroimaging, neuromodulation, and advanced mobile technologies in psychiatry practice. Additionally, we review the role of ML in molecular phenotyping and cross-species biomarker identification in precision psychiatry. We further discuss explainable AI (XAI) and causality testing in a closed-human-in-the-loop manner, and highlight the ML potential in multimedia information extraction and multimodal data fusion. Finally, we discuss conceptual and practical challenges in precision psychiatry and highlight ML opportunities in future research.

cs.LG↗

arXiv2022

Deformer: Towards Displacement Field Learning for Unsupervised Medical Image Registration

Recently, deep-learning-based approaches have been widely studied for deformable image registration task. However, most efforts directly map the composite image representation to spatial transformation through the convolutional neural network, ignoring its limited ability to capture spatial correspondence. On the other hand, Transformer can better characterize the spatial relationship with attention mechanism, its long-range dependency may be harmful to the registration task, where voxels with too large distances are unlikely to be corresponding pairs. In this study, we propose a novel Deformer module along with a multi-scale framework for the deformable image registration task. The Deformer module is designed to facilitate the mapping from image representation to spatial transformation by formulating the displacement vector prediction as the weighted summation of several bases. With the multi-scale framework to predict the displacement fields in a coarse-to-fine manner, superior performance can be achieved compared with traditional and learning-based approaches. Comprehensive experiments on two public datasets are conducted to demonstrate the effectiveness of the proposed Deformer module as well as the multi-scale framework.

eess.IV↗

arXiv2022

Dissecting Service Mesh Overheads

Service meshes play a central role in the modern application ecosystem by providing an easy and flexible way to connect different services that form a distributed application. However, because of the way they interpose on application traffic, they can substantially increase application latency and resource consumption. We develop a decompositional approach and a tool, called MeshInsight, to systematically characterize the overhead of service meshes and to help developers quantify overhead in deployment scenarios of interest. Using MeshInsight, we confirm that service meshes can have high overhead -- up to 185% higher latency and up to 92% more virtual CPU cores for our benchmark applications -- but the severity is intimately tied to how they are configured and the application workload. The primary contributors to overhead vary based on the configuration too. IPC (inter-process communication) and socket writes dominate when the service mesh operates as a TCP proxy, but protocol parsing dominates when it operates as an HTTP proxy. MeshInsight also enables us to study the end-to-end impact of optimizations to service meshes. We show that not all seemingly-promising optimizations lead to a notable overhead reduction in realistic settings.

cs.DC↗

arXiv2022

MAESTRO: Matched Speech Text Representations through Modality Matching

We present Maestro, a self-supervised training method to unify representations learnt from speech and text modalities. Self-supervised learning from speech signals aims to learn the latent structure inherent in the signal, while self-supervised learning from text attempts to capture lexical information. Learning aligned representations from unpaired speech and text sequences is a challenging task. Previous work either implicitly enforced the representations learnt from these two modalities to be aligned in the latent space through multitasking and parameter sharing or explicitly through conversion of modalities via speech synthesis. While the former suffers from interference between the two modalities, the latter introduces additional complexity. In this paper, we propose Maestro, a novel algorithm to learn unified representations from both these modalities simultaneously that can transfer to diverse downstream tasks such as Automated Speech Recognition (ASR) and Speech Translation (ST). Maestro learns unified representations through sequence alignment, duration prediction and matching embeddings in the learned space through an aligned masked-language model loss. We establish a new state-of-the-art (SOTA) on VoxPopuli multilingual ASR with a 8% relative reduction in Word Error Rate (WER), multidomain SpeechStew ASR (3.7% relative) and 21 languages to English multilingual ST on CoVoST 2 with an improvement of 2.8 BLEU averaged over 21 languages.

cs.CL↗

arXiv2022

Self-supervised Learning with Random-projection Quantizer for Speech Recognition

We present a simple and effective self-supervised learning approach for speech recognition. The approach learns a model to predict the masked speech signals, in the form of discrete labels generated with a random-projection quantizer. In particular the quantizer projects speech inputs with a randomly initialized matrix, and does a nearest-neighbor lookup in a randomly-initialized codebook. Neither the matrix nor the codebook is updated during self-supervised learning. Since the random-projection quantizer is not trained and is separated from the speech recognition model, the design makes the approach flexible and is compatible with universal speech recognition architecture. On LibriSpeech our approach achieves similar word-error-rates as previous work using self-supervised learning with non-streaming models, and provides lower word-error-rates and latency than wav2vec 2.0 and w2v-BERT with streaming models. On multilingual tasks the approach also provides significant improvement over wav2vec 2.0 and w2v-BERT.

cs.CL↗

arXiv2022

Leveraging unsupervised and weakly-supervised data to improve direct speech-to-speech translation

End-to-end speech-to-speech translation (S2ST) without relying on intermediate text representations is a rapidly emerging frontier of research. Recent works have demonstrated that the performance of such direct S2ST systems is approaching that of conventional cascade S2ST when trained on comparable datasets. However, in practice, the performance of direct S2ST is bounded by the availability of paired S2ST training data. In this work, we explore multiple approaches for leveraging much more widely available unsupervised and weakly-supervised speech and text data to improve the performance of direct S2ST based on Translatotron 2. With our most effective approaches, the average translation quality of direct S2ST on 21 language pairs on the CVSS-C corpus is improved by +13.6 BLEU (or +113% relatively), as compared to the previous state-of-the-art trained without additional data. The improvements on low-resource language are even more significant (+398% relatively on average). Our comparative studies suggest future research directions for S2ST and speech representation learning.

cs.CL↗

arXiv2022

Model-Based Disturbance Estimation for a Fiber-Reinforced Soft Manipulator using Orientation Sensing

For soft robots to work effectively in human-centered environments, they need to be able to estimate their state and external interactions based on (proprioceptive) sensors. Estimating disturbances allows a soft robot to perform desirable force control. Even in the case of rigid manipulators, force estimation at the end-effector is seen as a non-trivial problem. And indeed, other current approaches to address this challenge have shortcomings that prevent their general application. They are often based on simplified soft dynamic models, such as the ones relying on a piece-wise constant curvature (PCC) approximation or matched rigid-body models that do not represent enough details of the problem. Thus, the applications needed for complex human-robot interaction can not be built. Finite element methods (FEM) allow for predictions of soft robot dynamics in a more generic fashion. Here, using the soft robot modeling capabilities of the framework SOFA, we build a detailed FEM model of a multi-segment soft continuum robotic arm composed of compliant deformable materials and fiber-reinforced pressurized actuation chambers with a model for sensors that provide orientation output. This model is used to establish a state observer for the manipulator. Model parameters were calibrated to match imperfections of the manual fabrication process using physical experiments. We then solve a quadratic programming inverse dynamics problem to compute the components of external force that explain the pose error. Our experiments show an average force estimation error of around 1.2%. As the methods proposed are generic, these results are encouraging for the task of building soft robots exhibiting complex, reactive, sensor-based behavior that can be deployed in human-centered environments.

cs.RO↗

arXiv2022

LightHuBERT: Lightweight and Configurable Speech Representation Learning with Once-for-All Hidden-Unit BERT

Self-supervised speech representation learning has shown promising results in various speech processing tasks. However, the pre-trained models, e.g., HuBERT, are storage-intensive Transformers, limiting their scope of applications under low-resource settings. To this end, we propose LightHuBERT, a once-for-all Transformer compression framework, to find the desired architectures automatically by pruning structured parameters. More precisely, we create a Transformer-based supernet that is nested with thousands of weight-sharing subnets and design a two-stage distillation strategy to leverage the contextualized latent representations from HuBERT. Experiments on automatic speech recognition (ASR) and the SUPERB benchmark show the proposed LightHuBERT enables over $10^9$ architectures concerning the embedding dimension, attention dimension, head number, feed-forward network ratio, and network depth. LightHuBERT outperforms the original HuBERT on ASR and five SUPERB tasks with the HuBERT size, achieves comparable performance to the teacher model in most tasks with a reduction of 29% parameters, and obtains a $3.5\times$ compression ratio in three SUPERB tasks, e.g., automatic speaker verification, keyword spotting, and intent classification, with a slight accuracy loss. The code and pre-trained models are available at https://github.com/mechanicalsea/lighthubert.

eess.AS↗

arXiv2022

Structured Light with Redundancy Codes

Structured light (SL) systems acquire high-fidelity 3D geometry with active illumination projection. Conventional systems exhibit challenges when working in environments with strong ambient illumination, global illumination and cross-device interference. This paper proposes a general-purposed technique to improve the robustness of SL by projecting redundant optical signals in addition to the native SL patterns. In this way, projected signals become more distinguishable from errors. Thus the geometry information can be more easily recovered using simple signal processing and the ``coding gain" in performance is obtained. We propose three applications using our redundancy codes: (1) Self error-correction for SL imaging under strong ambient light, (2) Error detection for adaptive reconstruction under global illumination, and (3) Interference filtering with device-specific projection sequence encoding, especially for event camera-based SL and light curtain devices. We systematically analyze the design rules and signal processing algorithms in these applications. Corresponding hardware prototypes are built for evaluations on real-world complex scenes. Experimental results on the synthetic and real data demonstrate the significant performance improvements in SL systems with our redundancy codes.

cs.CV↗

arXiv2022

Spectral fluctuations in the interacting boson model

The energy dependence of the spectral fluctuations in the interacting boson model (IBM) and its connections to the mean-field structures have been analyzed through adopting two statistical measures, the nearest neighbor level spacing distribution $P(S)$ measuring the chaoticity (regularity) in energy spectra and the $Δ_3(L)$ statistics of Dyson and Metha measuring the spectral rigidity. Specifically, the statistical results as functions of the energy cutoff have been worked out for different dynamical situations including the U(5)-SU(3) and SU(3)-O(6) transitions as well as those near the AW arc of regularity. It is found that most of the changes in spectral fluctuations are triggered near the stationary points of the classical potential especially for the cases in the deformed region of the IBM phase diagram. The results thus justify the stationary point effects from the point of view of statistics. In addition, the approximate degeneracies in the $2^+$ spectrum on the AW arc is also revealed from the statistical calculations.

nucl-th↗

arXiv2022

Study of 20-inch PMTs dark count generated large pulses

The main goal of the JUNO experiment is to determine the neutrino mass ordering with a 20\,kt liquid-scintillator detector. The 20-inch PMT is crucial as one of JUNO key instruments to realize an excellent energy resolution of at least 3\,\% at 1\,MeV. The knowledge on PMT's characterisation and feature is critical for detector performance understanding. Large pulses from PMT dark count such as from flasher or others are one of the serious concerns for detector noise control. Focusing on the large pulses from 20-inch PMT dark count, this paper is trying to investigate the causes by measurements with a muon tagging system. It is found that the large pules of 20-inch PMT dark count is contributed mainly from muons hitting the PMT glass. A simulation is also realized and achieved a consistent understanding.

physics.ins-det↗

arXiv2022

ET White Paper: To Find the First Earth 2.0

We propose to develop a wide-field and ultra-high-precision photometric survey mission, temporarily named "Earth 2.0 (ET)". This mission is designed to measure, for the first time, the occurrence rate and the orbital distributions of Earth-sized planets. ET consists of seven 30cm telescopes, to be launched to the Earth-Sun's L2 point. Six of these are transit telescopes with a field of view of 500 square degrees. Staring in the direction that encompasses the original Kepler field for four continuous years, this monitoring will return tens of thousands of transiting planets, including the elusive Earth twins orbiting solar-type stars. The seventh telescope is a 30cm microlensing telescope that will monitor an area of 4 square degrees toward the galactic bulge. This, combined with simultaneous ground-based KMTNet observations, will measure masses for hundreds of long-period and free-floating planets. Together, the transit and the microlensing telescopes will revolutionize our understandings of terrestrial planets across a large swath of orbital distances and free space. In addition, the survey data will also facilitate studies in the fields of asteroseismology, Galactic archeology, time-domain sciences, and black holes in binaries.

astro-ph.IM↗

arXiv2022

Searching for Variable Stars in the Open Cluster NGC 2355 and Its Surrounding Region

We have investigated the variable stars in the field surrounding NGC 2355 based on the time-series photometric observation data. More than 3000 CCD frames were obtained in the V band spread over 13 nights with the Nanshan One-meter Wide-field Telescope. We have detected 88 variable stars, containing 72 new variable stars and 16 known variable stars. By analyzing these light curves, we classified the variable stars as follows: 26 eclipsing binaries, 52 pulsating stars, 4 rotating variables, and 6 unclear type variable stars for which their periods are much longer than the time baseline chosen. Employing Gaia DR2 parallax, kinematics, and photometry, the cluster membership of these variable stars were also analyzed for NGC 2355. In addition to the 11 variable members reported by Cantat-Gaudin et al. (2018), we identify 4 more variable member candidates located at the outer region of NGC 2355 and showed homogeneity in space positions and kinematic properties with the cluster members. The main physical parameters of NGC 2355 estimated from the two-color and color-magnitude diagrams are log(age/yr) = 8.9, E(B - V) = 0.24 mag, and [Fe/H] = - 0.07 dex.

astro-ph.SR↗

arXiv2022

Recent advances of defect-induced spin and valley polarized states in graphene

Electrons in graphene have fourfold spin and valley degeneracies owing to the unique bipartite honeycomb lattice and an extremely weak spin-orbit coupling, which can support a series of broken symmetry states. Atomic-scale defects in graphene are expected to lift these degenerate degrees of freedom at the nanoscale, and hence, lead to rich quantum states, highlighting promising directions for spintronics and valleytronics. In this article, we mainly review the recent scanning tunneling microscopy (STM) advances on the spin and/or valley polarized states induced by an individual atomic-scale defect in graphene, including a single-carbon vacancy, a nitrogen-atom dopant, and a hydrogen-atom chemisorption. Lastly, we give a perspective in this field.

cond-mat.mes-hall↗

arXiv2022

Characterization and manipulation of intervalley scattering induced by an individual monovacancy in graphene

Intervalley scattering involves microscopic processes that electrons are scattered by atomic-scale defects on nanometer length scales. Although central to our understanding of electronic properties of materials, direct characterization and manipulation of range and strength of the intervalley scattering induced by an individual atomic defect have so far been elusive. Using scanning tunneling microscope, we visualized and controlled intervalley scattering from an individual monovacancy in graphene. By directly imaging the affected range of intervalley scattering of the monovacancy, we demonstrated that it is inversely proportional to the energy, i.e., it is proportional to the wavelength of massless Dirac Fermions. A giant electron-hole asymmetry of the intervalley scattering is observed because that the monovacancy is charged. By further charging the monovacancy, the bended electronic potential around the monovacancy softened the scattering potential, which, consequently, suppressed the intervalley scattering of the monovacancy.

cond-mat.mes-hall↗

arXiv2022

Predicting Axillary Lymph Node Metastasis in Early Breast Cancer Using Deep Learning on Primary Tumor Biopsy Slides

Objectives: To develop and validate a deep learning (DL)-based primary tumor biopsy signature for predicting axillary lymph node (ALN) metastasis preoperatively in early breast cancer (EBC) patients with clinically negative ALN. Methods: A total of 1,058 EBC patients with pathologically confirmed ALN status were enrolled from May 2010 to August 2020. A DL core-needle biopsy (DL-CNB) model was built on the attention-based multiple instance-learning (AMIL) framework to predict ALN status utilizing the DL features, which were extracted from the cancer areas of digitized whole-slide images (WSIs) of breast CNB specimens annotated by two pathologists. Accuracy, sensitivity, specificity, receiver operating characteristic (ROC) curves, and areas under the ROC curve (AUCs) were analyzed to evaluate our model. Results: The best-performing DL-CNB model with VGG16_BN as the feature extractor achieved an AUC of 0.816 (95% confidence interval (CI): 0.758, 0.865) in predicting positive ALN metastasis in the independent test cohort. Furthermore, our model incorporating the clinical data, which was called DL-CNB+C, yielded the best accuracy of 0.831 (95%CI: 0.775, 0.878), especially for patients younger than 50 years (AUC: 0.918, 95%CI: 0.825, 0.971). The interpretation of DL-CNB model showed that the top signatures most predictive of ALN metastasis were characterized by the nucleus features including density ($p$ = 0.015), circumference ($p$ = 0.009), circularity ($p$ = 0.010), and orientation ($p$ = 0.012). Conclusion: Our study provides a novel DL-based biomarker on primary tumor CNB slides to predict the metastatic status of ALN preoperatively for patients with EBC. The codes and dataset are available at https://github.com/bupt-ai-cz/BALNMP

eess.IV↗

arXiv2022

Soft Retargeting Network for Click Through Rate Prediction

The study of user interest models has received a great deal of attention in click through rate (CTR) prediction recently. These models aim at capturing user interest from different perspectives, including user interest evolution, session interest, multiple interests, etc. In this paper, we focus on a new type of user interest, i.e., user retargeting interest. User retargeting interest is defined as user's click interest on target items the same as or similar to historical click items. We propose a novel soft retargeting network (SRN) to model this specific interest. Specifically, we first calculate the similarity between target item and each historical item with the help of graph embedding. Then we learn to aggregate the similarity weights to measure the extent of user's click interest on target item. Furthermore, we model the evolution of user retargeting interest. Experimental results on public datasets and industrial dataset demonstrate that our model achieves significant improvements over state-of-the-art models.

cs.IR↗

arXiv2022

PHEE: A phased hybrid evaluation-enhanced approach for identifying influential users in social networks

For the purpose of maximizing the spread of influence caused by a certain small number k of nodes in a social network, we are asked to find a k-subset of nodes (i.e., a seed set) with the best capacity to influence the nodes not in it. This problem of influence maximization (IM) has wide application, belongs to subset problems, and is NP-hard. To solve it, we should theoretically examine all seed sets and evaluate their influence spreads, which is time-consuming. Therefore, metaheuristic strategies are generally employed to gain a good seed set within a reasonable time. We observe that many algorithms for the IM problem only adopt a uniform mechanism in the whole solution search process, which lacks a response measure when the algorithm becomes trapped in a local optimum. To address this issue, we propose a phased hybrid evaluation-enhanced (PHEE) approach for IM, which utilizes two distinct search strategies to enhance the search of optimal solutions: a randomized range division evolutionary (RandRDE) algorithm to improve the solution quality, and a fast convergence strategy. Our approach is evaluated on 10 real-world social networks of different sizes and types. Experimental results demonstrate that our algorithm is efficient and obtains the best influence spread for all the datasets compared with three state-of-the-art algorithms, outperforms the time consuming CELF algorithm on four datasets, and performs worse than CELF on only two networks.

cs.SI↗