Searcharxiv⌕ Search

arXiv subjects

Yan Shen

Publications and source records attributed to Yan Shen.

48 records · Page 3Linked to original sources

A Bayesian Detect to Track System for Robust Visual Object Tracking and Semi-Supervised Model Learning

Object tracking is one of the fundamental problems in visual recognition tasks and has achieved significant improvements in recent years. The achievements often come with the price of enormous hardware consumption and expensive labor effort for consecutive labeling. A missing ingredient for robust tracking is achieving performance with minimal modification on network structure and semi-supervised learning intermittent labeled frames. In this paper, we ad-dress these problems in a Bayesian tracking and detection framework parameterized by neural network outputs. In our framework, the tracking and detection process is formulated in a probabilistic way as multi-objects dynamics and network detection uncertainties. With our formulation, we propose a particle filter-based approximate sampling algorithm for tracking object state estimation. Based on our particle filter inference algorithm, a semi-supervised learn-ing algorithm is utilized for learning tracking network on intermittent labeled frames by variational inference. In our experiments, we provide both mAP and probability-based detection measurements for comparison between our algorithm with non-Bayesian solutions. We also train a semi-supervised tracking network on M2Cai16-Tool-Locations Dataset and compare our results with supervised learning on fully labeled frames.

cs.CV↗

Confidence Intervals of Treatment Effects in Panel Data Models with Interactive Fixed Effects

We consider the construction of confidence intervals for treatment effects estimated using panel models with interactive fixed effects. We first use the factor-based matrix completion technique proposed by Bai and Ng (2021) to estimate the treatment effects, and then use bootstrap method to construct confidence intervals of the treatment effects for treated units at each post-treatment period. Our construction of confidence intervals requires neither specific distributional assumptions on the error terms nor large number of post-treatment periods. We also establish the validity of the proposed bootstrap procedure that these confidence intervals have asymptotically correct coverage probabilities. Simulation studies show that these confidence intervals have satisfactory finite sample performances, and empirical applications using classical datasets yield treatment effect estimates of similar magnitudes and reliable confidence intervals.

econ.EM↗

FedMM: Saddle Point Optimization for Federated Adversarial Domain Adaptation

Federated adversary domain adaptation is a unique distributed minimax training task due to the prevalence of label imbalance among clients, with each client only seeing a subset of the classes of labels required to train a global model. To tackle this problem, we propose a distributed minimax optimizer referred to as FedMM, designed specifically for the federated adversary domain adaptation problem. It works well even in the extreme case where each client has different label classes and some clients only have unsupervised tasks. We prove that FedMM ensures convergence to a stationary point with domain-shifted unsupervised data. On a variety of benchmark datasets, extensive experiments show that FedMM consistently achieves either significant communication savings or significant accuracy improvements over federated optimizers based on the gradient descent ascent (GDA) algorithm. When training from scratch, for example, it outperforms other GDA based federated average methods by around $20\%$ in accuracy over the same communication rounds; and it consistently outperforms when training from pre-trained models with an accuracy improvement from $5.4\%$ to $9\%$ for different networks.

cs.LG↗

An End-to-End learnable Flow Regularized Model for Brain Tumor Segmentation

Many segmentation tasks for biomedical images can be modeled as the minimization of an energy function and solved by a class of max-flow and min-cut optimization algorithms. However, the segmentation accuracy is sensitive to the contrasting of semantic features of different segmenting objects, as the traditional energy function usually uses hand-crafted features in their energy functions. To address these limitations, we propose to incorporate end-to-end trainable neural network features into the energy functions. Our deep neural network features are extracted from the down-sampling and up-sampling layers with skip-connections of a U-net. In the inference stage, the learned features are fed into the energy functions. And the segmentations are solved in a primal-dual form by ADMM solvers. In the training stage, we train our neural networks by optimizing the energy function in the primal form with regularizations on the min-cut and flow-conservation functions, which are derived from the optimal conditions in the dual form. We evaluate our methods, both qualitatively and quantitatively, in a brain tumor segmentation task. As the energy minimization model achieves a balance on sensitivity and smooth boundaries, we would show how our segmentation contours evolve actively through iterations as ensemble references for doctor diagnosis.

cs.CV↗

Scribble-based Hierarchical Weakly Supervised Learning for Brain Tumor Segmentation

The recent state-of-the-art deep learning methods have significantly improved brain tumor segmentation. However, fully supervised training requires a large amount of manually labeled masks, which is highly time-consuming and needs domain expertise. Weakly supervised learning with scribbles provides a good trade-off between model accuracy and the effort of manual labeling. However, for segmenting the hierarchical brain tumor structures, manually labeling scribbles for each substructure could still be demanding. In this paper, we use only two kinds of weak labels, i.e., scribbles on whole tumor and healthy brain tissue, and global labels for the presence of each substructure, to train a deep learning model to segment all the sub-regions. Specifically, we train two networks in two phases: first, we only use whole tumor scribbles to train a whole tumor (WT) segmentation network, which roughly recovers the WT mask of training data; then we cluster the WT region with the guide of global labels. The rough substructure segmentation from clustering is used as weak labels to train the second network. The dense CRF loss is used to refine the weakly supervised segmentation. We evaluate our approach on the BraTS2017 dataset and achieve competitive WT dice score as well as comparable scores on substructure segmentation compared to an upper bound when trained with fully annotated masks.

eess.IV↗

Equivalence of pth moment stability between stochastic differential delay equations and their numerical methods

In this paper, a general theorem on the equivalence of pth moment stability between stochastic differential delay equations (SDDEs) and their numerical methods is proved under the assumptions that the numerical methods are strongly convergent and have the bouneded $p$th moment in the finite time. The truncated Euler-Maruyama (EM) method is studied as an example to illustrate that the theorem indeed covers a large ranges of SDDEs. Alongside the investigation of the truncated EM method, the requirements on the step size of the method are significantly released compared with the work, where the method was initially proposed.

math.NA↗

Brain Tumor Segmentation on MRI with Missing Modalities

Brain Tumor Segmentation from magnetic resonance imaging (MRI) is a critical technique for early diagnosis. However, rather than having complete four modalities as in BraTS dataset, it is common to have missing modalities in clinical scenarios. We design a brain tumor segmentation algorithm that is robust to the absence of any modality. Our network includes a channel-independent encoding path and a feature-fusion decoding path. We use self-supervised training through channel dropout and also propose a novel domain adaptation method on feature maps to recover the information from the missing channel. Our results demonstrate that the quality of the segmentation depends on which modality is missing. Furthermore, we also discuss and visualize the contribution of each modality to the segmentation results. Their contributions are along well with the expert screening routine.

cs.CV↗

Dynamic Routing on Deep Neural Network for Thoracic Disease Classification and Sensitive Area Localization

We present and evaluate a new deep neural network architecture for automatic thoracic disease detection on chest X-rays. Deep neural networks have shown great success in a plethora of visual recognition tasks such as image classification and object detection by stacking multiple layers of convolutional neural networks (CNN) in a feed-forward manner. However, the performance gain by going deeper has reached bottlenecks as a result of the trade-off between model complexity and discrimination power. We address this problem by utilizing the recently developed routing-by agreement mechanism in our architecture. A novel characteristic of our network structure is that it extends routing to two types of layer connections (1) connection between feature maps in dense layers, (2) connection between primary capsules and prediction capsules in final classification layer. We show that our networks achieve comparable results with much fewer layers in the measurement of AUC score. We further show the combined benefits of model interpretability by generating Gradient-weighted Class Activation Mapping (Grad-CAM) for localization. We demonstrate our results on the NIH chestX-ray14 dataset that consists of 112,120 images on 30,805 unique patients including 14 kinds of lung diseases.

cs.CV↗

When Renewable Energy Meets Building Thermal Mass: A Real-time Load Management Scheme

We consider the optimal power management in renewable driven smart building MicroGrid under noise corrupted conditions as a stochastic optimization problem. We first propose our user satisfaction and electricity consumption balanced (USECB) profit model as the objective for optimal power management. We then cast the problem in noise corrupted conditions into the class of expectation maximizing in stochastic optimization problem with convex constraints. For this task, we design a Bregemen projection based mirror decent algorithm as an approximation solution to our stochastic optimization problem. Convergence and upper-bound of our algorithm with proof are also provided in our paper. We then conduct a broad type of experiment in our simulation to test the justification of our model as well as the effectiveness of our algorithm.

eess.SY↗

A general study on the volume dependence of spectral weights in lattice field theory

It has been suggested that the volume dependence of the spectral weight could be utilized to distinguish single and multi-particle states in Monte Carlo simulations. In a recent study using a solvable model, the Lee model, we found that this criteria is applicable only for stable particles and narrow resonances, not for the broad resonances. In this paper, the same question is addressed within the finite size formalism outlined by Lüscher. Using a quantum mechanical scattering model, the conclusion that was found in previous Lee model study is recovered. Then, following similar arguments as in Lüscher's, it is argued that the result is valid for a general massive quantum field theory under the same conditions as the Lüscher's formulae. Using the spectral weight function, a possibility of extracting resonance parameters is also pointed out.

hep-lat↗

Low-energy $D^{*+}D^0_1$ Scattering and the Resonance-like Structure $Z^+(4430)$

Low-energy scattering of $D^*$ and $D_1$ meson are studied using quenched lattice QCD with improved lattice actions on anisotropic lattices. The calculation is performed within Lüscher's finite-size formalism which establishes the relation between the scattering phase in the infinite volume and the exact energy level in the finite volume. The threshold scattering parameters, namely the scattering length $a_0$ and the effective range $r_0$, for the s-wave scattering in $J^P=0^-$ channel are extracted. After the chiral and continuum extrapolations, we obtain: $a_0=2.52(47)$fm and $r_0=0.7(1)$fm where the errors are purely statistical. Based on these results, we discuss the possibility of a shallow bound state for the two charmed mesons within the non-relativistic potential scattering model. It is argued that, albeit the interaction between the two charmed mesons being attractive, it is unlikely that they can form a shallow bound state in this channel. This calculation provides some useful information on the nature of the newly discovered resonance-like structure $Z^+(4430)$ by the Belle Collaboration.

hep-lat↗

Hadron Scattering in an Asymmetric Box

We propose to study hadron-hadron scattering using lattice QCD in an asymmetric box which allows one to access more non-degenerate low-momentum modes for a given volume. The conventional Lüscher's formula applicable in a symmetric box is modified accordingly. To illustrate the feasibility of this approach, pion-pion elastic scattering phase shifts in the I=2, J=0 channel are calculated within quenched approximation using improved gauge and Wilson fermion actions on anisotropic lattices in an asymmetric box. After the chiral and continuum extrapolation, we find that our quenched results for the scattering phase shifts in this channel are consistent with the experimental data when the three-momentum of the pion is below 300MeV. Agreement is also found when compared with previous theoretical results from lattice and other means. Moreover, with the usage of asymmetric volume, we are able to compute the scattering phases in the low-momentum range (pion three momentum less than about 350MeV in the center of mass frame) for over a dozen values of the pion three-momenta, much more than using the conventional symmetric box with comparable volume.

hep-lat↗