arXiv subjects

Ce Wang

Publications and source records attributed to Ce Wang.

At least 55 records · Page 3Linked to original sources

arXiv2022

Universal Segmentation of 33 Anatomies

In the paper, we present an approach for learning a single model that universally segments 33 anatomical structures, including vertebrae, pelvic bones, and abdominal organs. Our model building has to address the following challenges. Firstly, while it is ideal to learn such a model from a large-scale, fully-annotated dataset, it is practically hard to curate such a dataset. Thus, we resort to learn from a union of multiple datasets, with each dataset containing the images that are partially labeled. Secondly, along the line of partial labelling, we contribute an open-source, large-scale vertebra segmentation dataset for the benefit of spine analysis community, CTSpine1K, boasting over 1,000 3D volumes and over 11K annotated vertebrae. Thirdly, in a 3D medical image segmentation task, due to the limitation of GPU memory, we always train a model using cropped patches as inputs instead a whole 3D volume, which limits the amount of contextual information to be learned. To this, we propose a cross-patch transformer module to fuse more information in adjacent patches, which enlarges the aggregated receptive field for improved segmentation performance. This is especially important for segmenting, say, the elongated spine. Based on 7 partially labeled datasets that collectively contain about 2,800 3D volumes, we successfully learn such a universal model. Finally, we evaluate the universal model on multiple open-source datasets, proving that our model has a good generalization performance and can potentially serve as a solid foundation for downstream tasks.

eess.IV↗

arXiv2022

3D endoscopic depth estimation using 3D surface-aware constraints

Robotic-assisted surgery allows surgeons to conduct precise surgical operations with stereo vision and flexible motor control. However, the lack of 3D spatial perception limits situational awareness during procedures and hinders mastering surgical skills in the narrow abdominal space. Depth estimation, as a representative perception task, is typically defined as an image reconstruction problem. In this work, we show that depth estimation can be reformed from a 3D surface perspective. We propose a loss function for depth estimation that integrates the surface-aware constraints, leading to a faster and better convergence with the valid information from spatial information. In addition, camera parameters are incorporated into the training pipeline to increase the control and transparency of the depth estimation. We also integrate a specularity removal module to recover more buried image information. Quantitative experimental results on endoscopic datasets and user studies with medical professionals demonstrate the effectiveness of our method.

eess.IV↗

arXiv2021

Mott Insulator-Density Ordered Superfluid Transition and "Shamrock Transition" in a Frustrated Triangle Lattice

Density order is usually a consequence of the competition between long-range and short-range interactions. Here we report a density ordered superfluid emergent from a homogeneous Mott insulator due to the competition between frustrations and local interactions. This transition is found in a Bose-Hubbard model on a frustrated triangle lattice with an extra pairing term. Further, we find a quantum phase transition between two different density ordered superfluids, which is beyond the Landau-Ginzburg paradigm. Across this transition, a U(1) symmetry is emergent, while the symmetry in each density ordered superfluid is Z2*Z3. Because there emerges a shamrock-like degenerate ground state in parameter space, we call the transition "shamrock transition". Effective low energy theories are established for the two transitions mentioned above and we find their resemblance and differences with clock models.

cond-mat.quant-gas↗

arXiv2021

Noise Enhanced Neural Networks for Analytic Continuation

Analytic continuation maps imaginary-time Green's functions obtained by various theoretical/numerical methods to real-time response functions that can be directly compared with experiments. Analytic continuation is an important bridge between many-body theories and experiments but is also a challenging problem because such mappings are ill-conditioned. In this work, we develop a neural network-based method for this problem. The training data is generated either using synthetic Gaussian-type spectral functions or from exactly solvable models where the analytic continuation can be obtained analytically. Then, we applied the trained neural network to the testing data, either with synthetic noise or intrinsic noise in Monte Carlo simulations. We conclude that the best performance is always achieved when a proper amount of noise is added to the training data. Moreover, our method can successfully capture multi-peak structure in the resulting response function for the cases with the best performance. The method can be combined with Monte Carlo simulations to compare with experiments on real-time dynamics.

physics.comp-ph↗

arXiv2021

DuDoTrans: Dual-Domain Transformer Provides More Attention for Sinogram Restoration in Sparse-View CT Reconstruction

While Computed Tomography (CT) reconstruction from X-ray sinograms is necessary for clinical diagnosis, iodine radiation in the imaging process induces irreversible injury, thereby driving researchers to study sparse-view CT reconstruction, that is, recovering a high-quality CT image from a sparse set of sinogram views. Iterative models are proposed to alleviate the appeared artifacts in sparse-view CT images, but the computation cost is too expensive. Then deep-learning-based methods have gained prevalence due to the excellent performances and lower computation. However, these methods ignore the mismatch between the CNN's \textbf{local} feature extraction capability and the sinogram's \textbf{global} characteristics. To overcome the problem, we propose \textbf{Du}al-\textbf{Do}main \textbf{Trans}former (\textbf{DuDoTrans}) to simultaneously restore informative sinograms via the long-range dependency modeling capability of Transformer and reconstruct CT image with both the enhanced and raw sinograms. With such a novel design, reconstruction performance on the NIH-AAPM dataset and COVID-19 dataset experimentally confirms the effectiveness and generalizability of DuDoTrans with fewer involved parameters. Extensive experiments also demonstrate its robustness with different noise-level scenarios for sparse-view CT reconstruction. The code and models are publicly available at https://github.com/DuDoTrans/CODE

eess.IV↗

arXiv2021

Improving Generalizability in Limited-Angle CT Reconstruction with Sinogram Extrapolation

Computed tomography (CT) reconstruction from X-ray projections acquired within a limited angle range is challenging, especially when the angle range is extremely small. Both analytical and iterative models need more projections for effective modeling. Deep learning methods have gained prevalence due to their excellent reconstruction performances, but such success is mainly limited within the same dataset and does not generalize across datasets with different distributions. Hereby we propose ExtraPolationNetwork for limited-angle CT reconstruction via the introduction of a sinogram extrapolation module, which is theoretically justified. The module complements extra sinogram information and boots model generalizability. Extensive experimental results show that our reconstruction model achieves state-of-the-art performance on NIH-AAPM dataset, similar to existing approaches. More importantly, we show that using such a sinogram extrapolation module significantly improves the generalization capability of the model on unseen datasets (e.g., COVID-19 and LIDC datasets) when compared to existing approaches.

eess.IV↗

arXiv2021

Complex contact interaction for systems with short-range two-body loss

Contact interaction is a fundamental concept that appears in various areas of physics. It simplifies physical models by replacing the detailed short-range interaction with a zero-range contact potential which reproduces the same low-energy scattering parameter, i.e. the $s$-wave scattering length. In this work, we generalize this concept to a system with short-range two-body loss. We show that the short-range two-body loss can effectively be replaced by a zero-range complex contact potential with proper regularization characterized by a complex scattering length. We develop appropriate ways to regularize this potential in the Lindblad master equation and apply them to the dynamic problem of Bose-Einstein condensate with weak interaction and two-body loss.

cond-mat.quant-gas↗

arXiv2021

Recursive Training for Zero-Shot Semantic Segmentation

General purpose semantic segmentation relies on a backbone CNN network to extract discriminative features that help classify each image pixel into a 'seen' object class (ie., the object classes available during training) or a background class. Zero-shot semantic segmentation is a challenging task that requires a computer vision model to identify image pixels belonging to an object class which it has never seen before. Equipping a general purpose semantic segmentation model to separate image pixels of 'unseen' classes from the background remains an open challenge. Some recent models have approached this problem by fine-tuning the final pixel classification layer of a semantic segmentation model for a Zero-Shot setting, but struggle to learn discriminative features due to the lack of supervision. We propose a recursive training scheme to supervise the retraining of a semantic segmentation model for a zero-shot setting using a pseudo-feature representation. To this end, we propose a Zero-Shot Maximum Mean Discrepancy (ZS-MMD) loss that weighs high confidence outputs of the pixel classification layer as a pseudo-feature representation, and feeds it back to the generator. By closing-the-loop on the generator end, we provide supervision during retraining that in turn helps the model learn a more discriminative feature representation for 'unseen' classes. We show that using our recursive training and ZS-MMD loss, our proposed model achieves state-of-the-art performance on the Pascal-VOC 2012 dataset and Pascal-Context dataset.

cs.CV↗

arXiv2021

Measuring Bayesian Robustness Using Rényi Divergence

This paper deals with measuring the Bayesian robustness of classes of contaminated priors. Two different classes of priors in the neighborhood of the elicited prior are considered. The first one is the well-known $ε$-contaminated class, while the second one is the geometric mixing class. The proposed measure of robustness is based on computing the curvature of Rényi divergence between posterior distributions. Examples are used to illustrate the results by using simulated and real data sets.

math.ST↗

arXiv2020

Machine Learning Identification of Impurities in the STM Images

In this work we train a neural network to identify impurities in the experimental images obtained by the scanning tunneling microscope measurements. The neural network is first trained with large number of simulated data and then the trained neural network is applied to identify a set of experimental images taken at different voltages. We use the convolutional neural network to extract features from the images and also implement the attention mechanism to capture the correlations between images taken at different voltages. We note that the simulated data can capture the universal Friedel oscillation but cannot properly describe the non-universal physics short-range physics nearby an impurity, as well as noises in the experimental data. And we emphasize that the key of this approach is to properly deal these differences between simulated data and experimental data. Here we show that even by including uncorrelated white noises in the simulated data, the performance of neural network on experimental data can be significantly improved. To prevent the neural network from learning unphysical short-range physics, we also develop another method to evaluate the confidence of the neural network prediction on experimental data and to add this confidence measure into the loss function. We show that adding such an extra loss function can also improve the performance on experimental data. Our research can inspire future similar applications of machine learning on experimental data analysis.

cond-mat.str-el↗

arXiv2020

Higher-Dimensional Open Quantum Walk Constructed from Quantum Bernoulli Noises

Quantum Bernoulli noises are annihilation and creation operators acting on Bernoulli functionals, which satisfy the canonical anti-commutation relations (CAR) in equal-time. In this paper, we use quantum Bernoulli noises to introduce a model of open quantum walk on the $d$-dimensional integer lattice $\mathbb{Z}^d$ for a general positive integer $d\geq 2$, which we call the $d$-dimensional open QBN walk. We obtain a quantum channel representation of the $d$-dimensional open QBN walk, and find that it admits the ``separability-preserving'' property. We prove that, for a wide range of choices of its initial state, the $d$-dimensional open QBN walk has a limit probability distribution of $d$-dimensional Gauss type. Finally we unveil links between the $d$-dimensional open QBN walk and the unitary quantum walk recently introduced in [Ce Wang and Caishi Wang, Higher-dimensional quantum walk in terms of quantum Bernoulli noises, Entropy 2020, 22, 504].

math.PR↗

arXiv2020

First image then video: A two-stage network for spatiotemporal video denoising

Video denoising is to remove noise from noise-corrupted data, thus recovering true signals via spatiotemporal processing. Existing approaches for spatiotemporal video denoising tend to suffer from motion blur artifacts, that is, the boundary of a moving object tends to appear blurry especially when the object undergoes a fast motion, causing optical flow calculation to break down. In this paper, we address this challenge by designing a first-image-then-video two-stage denoising neural network, consisting of an image denoising module for spatially reducing intra-frame noise followed by a regular spatiotemporal video denoising module. The intuition is simple yet powerful and effective: the first stage of image denoising effectively reduces the noise level and, therefore, allows the second stage of spatiotemporal denoising for better modeling and learning everywhere, including along the moving object boundaries. This two-stage network, when trained in an end-to-end fashion, yields the state-of-the-art performances on the video denoising benchmark Vimeo90K dataset in terms of both denoising quality and computation. It also enables an unsupervised approach that achieves comparable performance to existing supervised approaches.

cs.CV↗

arXiv2019

Linking invariant for the quench dynamics of a two-dimensional two-band Chern insulator

We discuss the topological invariant in the (2+1)-dimensional quench dynamics of a two-dimensional two-band Chern insulator starting from a topological initial state (i.e., with a nonzero Chern number $c_i$), evolved by a post-quench Hamiltonian (with Chern number $c_f$). In contrast to the process with $c_i=0$ studied in previous works, this process cannot be characterized by the Hopf invariant that is described by the sphere homotopy group $π_3(S^2)=\mathbb{Z}$. It is possible, however, to calculate a variant of the Chern-Simons integral with a complementary part to cancel the Chern number of the initial spin configuration, which at the same time does not affect the (2+1)-dimensional topology. We show that the modified Chern-Simons integral gives rise to a topological invariant of this quench process, i.e., the linking invariant in the $\mathbb{Z}_{2c_i}$ class: $ν= (c_f - c_i) \mod (2c_i)$. We give concrete examples to illustrate this result and also show the detailed deduction to get this linking invariant.

cond-mat.quant-gas↗

arXiv2019

Emergent Quantum Mechanics in an Introspective Machine Learning Architecture

Can physical concepts and laws emerge in a neural network as it learns to predict the observation data of physical systems? As a benchmark and a proof-of-principle study of this possibility, here we show an introspective learning architecture that can automatically develop the concept of the quantum wave function and discover the Schrödinger equation from simulated experimental data of the potential-to-density mappings of a quantum particle. This introspective learning architecture contains a machine translator to perform the potential to density mapping, and a knowledge distiller auto-encoder to extract the essential information and its update law from the hidden states of the translator, which turns out to be the quantum wave function and the Schrödinger equation. We envision that our introspective learning architecture can enable machine learning to discover new physics in the future.

cond-mat.dis-nn↗

arXiv2019

Label-Removed Generative Adversarial Networks Incorporating with K-Means

Generative Adversarial Networks (GANs) have achieved great success in generating realistic images. Most of these are conditional models, although acquisition of class labels is expensive and time-consuming in practice. To reduce the dependence on labeled data, we propose an un-conditional generative adversarial model, called K-Means-GAN (KM-GAN), which incorporates the idea of updating centers in K-Means into GANs. Specifically, we redesign the framework of GANs by applying K-Means on the features extracted from the discriminator. With obtained labels from K-Means, we propose new objective functions from the perspective of deep metric learning (DML). Distinct from previous works, the discriminator is treated as a feature extractor rather than a classifier in KM-GAN, meanwhile utilization of K-Means makes features of the discriminator more representative. Experiments are conducted on various datasets, such as MNIST, Fashion-10, CIFAR-10 and CelebA, and show that the quality of samples generated by KM-GAN is comparable to some conditional generative adversarial models.

cs.LG↗

arXiv2018

Machine Learning of Frustrated Classical Spin Models. II. Kernel Principal Component Analysis

In this work we apply the principal component analysis (PCA) method with kernel trick to study classification of phases and phase transition in classical XY models in frustrated lattices. Comparing to our previous work with linear PCA method, the kernel PCA can capture non-linear function. In this case, the Z2 chiral order of classical spins in these lattices are indeed a non-linear function of the input spin configurations. In addition to the principal component revealed by linear PCA, the kernel PCA can find out two more principal components using data generated by Monte Carlo simulation at various temperatures at input. One of them relates to the strength of the U(1) order parameter and the other directly manifests the chiral order parameter that characterizes the Z2 symmetry breaking. For a temperature resolved study, the temperature dependence of the principal eigenvalue associated with the Z2 symmetry breaking clearly shows a second order phase transition behavior.

cond-mat.stat-mech↗

arXiv2017

Unsupervised Learning of Frustrated Classical Spin Models I: Principle Component Analysis

This work aims at the goal whether the artificial intelligence can recognize phase transition without the prior human knowledge. If this becomes successful, it can be applied to, for instance, analyze data from quantum simulation of unsolved physical models. Toward this goal, we first need to apply the machine learning algorithm to well-understood models and see whether the outputs are consistent with our prior knowledge, which serves as the benchmark of this approach. In this work, we feed the compute with data generated by the classical Monte Carlo simulation for the XY model in frustrated triangular and union jack lattices, which has two order parameters and exhibits two phase transitions. We show that the outputs of the principle component analysis agree very well with our understanding of different orders in different phases, and the temperature dependences of the major components detect the nature and the locations of the phase transitions. Our work offers promise for using machine learning techniques to study sophisticated statistical models, and our results can be further improved by using principle component analysis with kernel tricks and the neural network method.

cond-mat.stat-mech↗

arXiv2016

Measuring Topological Number of a Chern-Insulator from Quench Dynamics

In this letter we show how the topological number of a static Hamiltonian can be measured from a dynamical quench process. We focus on a two-band Chern insulator in two-dimension, for instance, the Haldane model, whose dynamical process can be described by a mapping from the $[k_x,k_y,t]$ space to the Bloch sphere, characterized by the Hopf invariant. Such a mapping has been constructed experimentally by measurements in cold atom systems. We show that, taking any two constant vectors on the Bloch sphere, their inverse images of this mapping are two trajectories in the $[k_x,k_y,t]$ space, and the linking number of these two trajectories exactly equals to the Chern number of the static Hamiltonian. Applying this result to a recent experiment from the Hamburg group, we show that the linking number of the trajectories of the phase vortices determines the phase boundary of the static Hamiltonian.

cond-mat.quant-gas↗