arXiv subjects

Lin Yang

Publications and source records attributed to Lin Yang.

At least 163 records · Page 9Linked to original sources

arXiv2022

Evaluating and Improving the Depth Accuracy of Kinect for Windows v2

Microsoft Kinect sensor has been widely used in many applications since the launch of its first version. Recently, Microsoft released a new version of Kinect sensor with improved hardware. However, the accuracy assessment of the sensor remains to be answered. In this paper, we measure the depth accuracy of the newly released Kinect v2 depth sensor, and obtain a cone model to illustrate its accuracy distribution. We then evaluate the variance of the captured depth values by depth entropy. In addition, we propose a trilateration method to improve the depth accuracy with multiple Kinects simultaneously. The experimental results are provided to ascertain the proposed model and method.

cs.MM↗

arXiv2022

3-D Markerless Tracking of Human Gait by Geometric Trilateration of Multiple Kinects

In this paper, we develop an integrated markerless gait tracking system with three Kinect v2 sensors. A geometric principle-based trilateration method is proposed for optimizing the accuracy of the measured gait data. To tackle the data synchronization problem among the Kinect clients and the server, a synchronization mechanism based on NTP (Network Time Protocol) is designed for synchronizing the server and Kinect clients' clocks. Furthermore, a time schedule is designed for timing each Kinect client's data transmission. In the experiment, participants are asked to perform a 60 s walk while the proposed tracking system obtains the participant's gait data. Six joints (including left hip, right hip, left knee, right knee, left ankle and right ankle) of the participants are tracked where the obtained gait data are described as 6000 {movements} of joint positions (1000 {movements} for each joint). The results show that the trilateration tracking result by the three Kinect sensors has a much higher accuracy compared with the accuracy measured by a single Kinect sensor. Within a randomly sampled time period (67.726 s in the experiment), 98.37% of the frames generated by the gait tracking system have timing errors less than 1 ms, which is much better than the default NTP service embedded in the Windows 8.1 operating system. The accuracy of the proposed system is quantitatively evaluated and verified by a comparison with a commercial medical system (Delsys Trigno Smart Sensor System).

cs.MM↗

arXiv2022

From Local to Global: Spectral-Inspired Graph Neural Networks

Graph Neural Networks (GNNs) are powerful deep learning methods for Non-Euclidean data. Popular GNNs are message-passing algorithms (MPNNs) that aggregate and combine signals in a local graph neighborhood. However, shallow MPNNs tend to miss long-range signals and perform poorly on some heterophilous graphs, while deep MPNNs can suffer from issues like over-smoothing or over-squashing. To mitigate such issues, existing works typically borrow normalization techniques from training neural networks on Euclidean data or modify the graph structures. Yet these approaches are not well-understood theoretically and could increase the overall computational complexity. In this work, we draw inspirations from spectral graph embedding and propose $\texttt{PowerEmbed}$ -- a simple layer-wise normalization technique to boost MPNNs. We show $\texttt{PowerEmbed}$ can provably express the top-$k$ leading eigenvectors of the graph operator, which prevents over-smoothing and is agnostic to the graph topology; meanwhile, it produces a list of representations ranging from local features to global signals, which avoids over-squashing. We apply $\texttt{PowerEmbed}$ in a wide range of simulated and real graphs and demonstrate its competitive performance, particularly for heterophilous graphs.

stat.ML↗

arXiv2022

GARNet: Global-Aware Multi-View 3D Reconstruction Network and the Cost-Performance Tradeoff

Deep learning technology has made great progress in multi-view 3D reconstruction tasks. At present, most mainstream solutions establish the mapping between views and shape of an object by assembling the networks of 2D encoder and 3D decoder as the basic structure while they adopt different approaches to obtain aggregation of features from several views. Among them, the methods using attention-based fusion perform better and more stable than the others, however, they still have an obvious shortcoming -- the strong independence of each view during predicting the weights for merging leads to a lack of adaption of the global state. In this paper, we propose a global-aware attention-based fusion approach that builds the correlation between each branch and the global to provide a comprehensive foundation for weights inference. In order to enhance the ability of the network, we introduce a novel loss function to supervise the shape overall and propose a dynamic two-stage training strategy that can effectively adapt to all reconstructors with attention-based fusion. Experiments on ShapeNet verify that our method outperforms existing SOTA methods while the amount of parameters is far less than the same type of algorithm, Pix2Vox++. Furthermore, we propose a view-reduction method based on maximizing diversity and discuss the cost-performance tradeoff of our model to achieve a better performance when facing heavy input amount and limited computational cost.

cs.CV↗

arXiv2022

Constraining Model Uncertainty in Plasma Equation-of-State Models with a Physics-Constrained Gaussian Process

Equation-of-state (EOS) models underpin numerical simulations at the core of research in high energy density physics, inertial confinement fusion, laboratory astrophysics, and elsewhere. In these applications EOS models are needed that span ranges of thermodynamic variables that far exceed the ranges where data are available, making uncertainty quantification (UQ) of EOS models a significant concern. Model uncertainty, arising from the choice of functional form assumed for the EOS, is a major challenge to UQ studies for EOS that is usually neglected in favor of parameteric and data uncertainties which are easier to capture without violating the physical constraints on EOSs. In this work we introduce a new statistical EOS construction that naturally captures model uncertainty while automatically obeying the thermodynamic consistency constraint. We apply the model to existing data for $B_4C$\ to place an upper bound on the uncertainty in the EOS and Hugoniot, and show that the neglect of thermodynamic constraints overestimates the uncertainty by factors of several when data are available and underestimates when extrapolating to regions where they are not. We discuss extensions to this approach, and the role of GP-based models in accelerating simulation and experimental studies, defining portable uncertainty-aware EOS tables, and enabling uncertainty-aware downstream tasks.

physics.data-an↗

arXiv2022

LeVoice ASR Systems for the ISCSLP 2022 Intelligent Cockpit Speech Recognition Challenge

This paper describes LeVoice automatic speech recognition systems to track2 of intelligent cockpit speech recognition challenge 2022. Track2 is a speech recognition task without limits on the scope of model size. Our main points include deep learning based speech enhancement, text-to-speech based speech generation, training data augmentation via various techniques and speech recognition model fusion. We compared and fused the hybrid architecture and two kinds of end-to-end architecture. For end-to-end modeling, we used models based on connectionist temporal classification/attention-based encoder-decoder architecture and recurrent neural network transducer/attention-based encoder-decoder architecture. The performance of these models is evaluated with an additional language model to improve word error rates. As a result, our system achieved 10.2\% character error rate on the challenge test set data and ranked third place among the submitted systems in the challenge.

eess.AS↗

arXiv2022

Unsupervised Dense Nuclei Detection and Segmentation with Prior Self-activation Map For Histology Images

The success of supervised deep learning models in medical image segmentation relies on detailed annotations. However, labor-intensive manual labeling is costly and inefficient, especially in dense object segmentation. To this end, we propose a self-supervised learning based approach with a Prior Self-activation Module (PSM) that generates self-activation maps from the input images to avoid labeling costs and further produce pseudo masks for the downstream task. To be specific, we firstly train a neural network using self-supervised learning and utilize the gradient information in the shallow layers of the network to generate self-activation maps. Afterwards, a semantic-guided generator is then introduced as a pipeline to transform visual representations from PSM to pixel-level semantic pseudo masks for downstream tasks. Furthermore, a two-stage training module, consisting of a nuclei detection network and a nuclei segmentation network, is adopted to achieve the final segmentation. Experimental results show the effectiveness on two public pathological datasets. Compared with other fully-supervised and weakly-supervised methods, our method can achieve competitive performance without any manual annotations.

cs.CV↗

arXiv2022

TC-SKNet with GridMask for Low-complexity Classification of Acoustic scene

Convolution neural networks (CNNs) have good performance in low-complexity classification tasks such as acoustic scene classifications (ASCs). However, there are few studies on the relationship between the length of target speech and the size of the convolution kernels. In this paper, we combine Selective Kernel Network with Temporal-Convolution (TC-SKNet) to adjust the receptive field of convolution kernels to solve the problem of variable length of target voice while keeping low-complexity. GridMask is a data augmentation strategy by masking part of the raw data or feature area. It can enhance the generalization of the model as the role of dropout. In our experiments, the performance gain brought by GridMask is stronger than spectrum augmentation in ASCs. Finally, we adopt AutoML to search best structure of TC-SKNet and hyperparameters of GridMask for improving the classification performance. As a result, a peak accuracy of 59.87% TC-SKNet is equivalent to that of SOTA, but the parameters only use 20.9 K.

cs.SD↗

arXiv2022

A Novel Physics-Regularized Interpretable Machine Learning Model for Grain Growth

Experimental grain growth observations often deviate from grain growth simulations, revealing that the governing rules for grain boundary motion are not fully understood. A novel deep learning model was developed to capture grain growth behavior from training data without making assumptions about the underlying physics. The Physics-Regularized Interpretable Machine Learning Microstructure Evolution (PRIMME) model consists of a multi-layer neural network that predicts the likelihood of a point changing to a neighboring grain. Here, we demonstrate PRIMME's ability to replicate two-dimensional normal grain growth by training it with Monte Carlo Potts simulations. The trained PRIMME model's grain growth predictions in several test cases show good agreement with analytical models, phase-field simulations, Monte Carlo Potts simulations, and results from the literature. Additionally, PRIMME's adaptability to investigate irregular grain growth behavior is shown. Important aspects of PRIMME like interpretability, regularization, extrapolation, and overfitting are also discussed.

physics.comp-ph↗

arXiv2022

Modelling matrix time series via a tensor CP-decomposition

We consider to model matrix time series based on a tensor CP-decomposition. Instead of using an iterative algorithm which is the standard practice for estimating CP-decompositions, we propose a new and one-pass estimation procedure based on a generalized eigenanalysis constructed from the serial dependence structure of the underlying process. To overcome the intricacy of solving a rank-reduced generalized eigenequation, we propose a further refined approach which projects it into a lower-dimensional full-ranked eigenequation. This refined method improves significantly the finite-sample performance of the estimation. The asymptotic theory has been established under a general setting without the stationarity. It shows, for example, that all the component coefficient vectors in the CP-decomposition are estimated consistently with certain convergence rates. The proposed model and the estimation method are also illustrated with both simulated and real data; showing effective dimension-reduction in modelling and forecasting matrix time series.

stat.ME↗

arXiv2022

Adaptive Decision Making at the Intersection for Autonomous Vehicles Based on Skill Discovery

In urban environments, the complex and uncertain intersection scenarios are challenging for autonomous driving. To ensure safety, it is crucial to develop an adaptive decision making system that can handle the interaction with other vehicles. Manually designed model-based methods are reliable in common scenarios. But in uncertain environments, they are not reliable, so learning-based methods are proposed, especially reinforcement learning (RL) methods. However, current RL methods need retraining when the scenarios change. In other words, current RL methods cannot reuse accumulated knowledge. They forget learned knowledge when new scenarios are given. To solve this problem, we propose a hierarchical framework that can autonomously accumulate and reuse knowledge. The proposed method combines the idea of motion primitives (MPs) with hierarchical reinforcement learning (HRL). It decomposes complex problems into multiple basic subtasks to reduce the difficulty. The proposed method and other baseline methods are tested in a challenging intersection scenario based on the CARLA simulator. The intersection scenario contains three different subtasks that can reflect the complexity and uncertainty of real traffic flow. After offline learning and testing, the proposed method is proved to have the best performance among all methods.

cs.RO↗

arXiv2022

ChrSNet: Chromosome Straightening using Self-attention Guided Networks

Karyotyping is an important procedure to assess the possible existence of chromosomal abnormalities. However, because of the non-rigid nature, chromosomes are usually heavily curved in microscopic images and such deformed shapes hinder the chromosome analysis for cytogeneticists. In this paper, we present a self-attention guided framework to erase the curvature of chromosomes. The proposed framework extracts spatial information and local textures to preserve banding patterns in a regression module. With complementary information from the bent chromosome, a refinement module is designed to further improve fine details. In addition, we propose two dedicated geometric constraints to maintain the length and restore the distortion of chromosomes. To train our framework, we create a synthetic dataset where curved chromosomes are generated from the real-world straight chromosomes by grid-deformation. Quantitative and qualitative experiments are conducted on synthetic and real-world data. Experimental results show that our proposed method can effectively straighten bent chromosomes while keeping banding details and length.

cs.CV↗

arXiv2022

End-to-end cell recognition by point annotation

Reliable quantitative analysis of immunohistochemical staining images requires accurate and robust cell detection and classification. Recent weakly-supervised methods usually estimate probability density maps for cell recognition. However, in dense cell scenarios, their performance can be limited by pre- and post-processing as it is impossible to find a universal parameter setting. In this paper, we introduce an end-to-end framework that applies direct regression and classification for preset anchor points. Specifically, we propose a pyramidal feature aggregation strategy to combine low-level features and high-level semantics simultaneously, which provides accurate cell recognition for our purely point-based model. In addition, an optimized cost function is designed to adapt our multi-task learning framework by matching ground truth and predicted points. The experimental results demonstrate the superior accuracy and efficiency of the proposed method, which reveals the high potentiality in assisting pathologist assessments.

cs.CV↗

arXiv2022

Benchmarking the Robustness of Deep Neural Networks to Common Corruptions in Digital Pathology

When designing a diagnostic model for a clinical application, it is crucial to guarantee the robustness of the model with respect to a wide range of image corruptions. Herein, an easy-to-use benchmark is established to evaluate how deep neural networks perform on corrupted pathology images. Specifically, corrupted images are generated by injecting nine types of common corruptions into validation images. Besides, two classification and one ranking metrics are designed to evaluate the prediction and confidence performance under corruption. Evaluated on two resulting benchmark datasets, we find that (1) a variety of deep neural network models suffer from a significant accuracy decrease (double the error on clean images) and the unreliable confidence estimation on corrupted images; (2) A low correlation between the validation and test errors while replacing the validation set with our benchmark can increase the correlation. Our codes are available on https://github.com/superjamessyx/robustness_benchmark.

cs.CV↗

arXiv2022

Low-Latency Online Speaker Diarization with Graph-Based Label Generation

This paper introduces an online speaker diarization system that can handle long-time audio with low latency. We enable Agglomerative Hierarchy Clustering (AHC) to work in an online fashion by introducing a label matching algorithm. This algorithm solves the inconsistency between output labels and hidden labels that are generated each turn. To ensure the low latency in the online setting, we introduce a variant of AHC, namely chkpt-AHC, to cluster the speakers. In addition, we propose a speaker embedding graph to exploit a graph-based re-clustering method, further improving the performance. In the experiment, we evaluate our systems on both DIHARD3 and VoxConverse datasets. The experimental results show that our proposed online systems have better performance than our baseline online system and have comparable performance to our offline systems. We find out that the framework combining the chkpt-AHC method and the label matching algorithm works well in the online setting. Moreover, the chkpt-AHC method greatly reduces the time cost, while the graph-based re-clustering method helps improve the performance.

eess.AS↗

arXiv2022

Model-based Offline Imitation Learning with Non-expert Data

Although Behavioral Cloning (BC) in theory suffers compounding errors, its scalability and simplicity still makes it an attractive imitation learning algorithm. In contrast, imitation approaches with adversarial training typically does not share the same problem, but necessitates interactions with the environment. Meanwhile, most imitation learning methods only utilises optimal datasets, which could be significantly more expensive to obtain than its suboptimal counterpart. A question that arises is, can we utilise the suboptimal dataset in a principled manner, which otherwise would have been idle? We propose a scalable model-based offline imitation learning algorithmic framework that leverages datasets collected by both suboptimal and optimal policies, and show that its worst case suboptimality becomes linear in the time horizon with respect to the expert samples. We empirically validate our theoretical results and show that the proposed method \textit{always} outperforms BC in the low data regime on simulated continuous control domains

cs.LG↗

arXiv2022

Discovery-and-Selection: Towards Optimal Multiple Instance Learning for Weakly Supervised Object Detection

Weakly supervised object detection (WSOD) is a challenging task that requires simultaneously learn object classifiers and estimate object locations under the supervision of image category labels. A major line of WSOD methods roots in multiple instance learning which regards images as bags of instances and selects positive instances from each bag to learn the detector. However, a grand challenge emerges when the detector inclines to converge to discriminative parts of objects rather than the whole objects. In this paper, under the hypothesis that optimal solutions are included in local minima, we propose a discovery-and-selection approach fused with multiple instance learning (DS-MIL), which finds rich local minima and select optimal solution from multiple local minima. To implement DS-MIL, an attention module is proposed so that more context information can be captured by feature maps and more valuable proposals can be collected during training. With proposal candidates, a selection module is proposed to select informative instances for object detector. Experimental results on commonly used benchmarks show that our proposed DS-MIL approach can consistently improve the baselines, reporting state-of-the-art performance.

cs.CV↗

arXiv2022

Invariant Content Synergistic Learning for Domain Generalization of Medical Image Segmentation

While achieving remarkable success for medical image segmentation, deep convolution neural networks (DCNNs) often fail to maintain their robustness when confronting test data with the novel distribution. To address such a drawback, the inductive bias of DCNNs is recently well-recognized. Specifically, DCNNs exhibit an inductive bias towards image style (e.g., superficial texture) rather than invariant content (e.g., object shapes). In this paper, we propose a method, named Invariant Content Synergistic Learning (ICSL), to improve the generalization ability of DCNNs on unseen datasets by controlling the inductive bias. First, ICSL mixes the style of training instances to perturb the training distribution. That is to say, more diverse domains or styles would be made available for training DCNNs. Based on the perturbed distribution, we carefully design a dual-branches invariant content synergistic learning strategy to prevent style-biased predictions and focus more on the invariant content. Extensive experimental results on two typical medical image segmentation tasks show that our approach performs better than state-of-the-art domain generalization methods.

eess.IV↗