Searcharxiv⌕ Search

arXiv subjects

Hong Liu

Publications and source records attributed to Hong Liu.

At least 271 records · Page 15Linked to original sources
arXiv2023

GATOR: Graph-Aware Transformer with Motion-Disentangled Regression for Human Mesh Recovery from a 2D Pose

3D human mesh recovery from a 2D pose plays an important role in various applications. However, it is hard for existing methods to simultaneously capture the multiple relations during the evolution from skeleton to mesh, including joint-joint, joint-vertex and vertex-vertex relations, which often leads to implausible results. To address this issue, we propose a novel solution, called GATOR, that contains an encoder of Graph-Aware Transformer (GAT) and a decoder with Motion-Disentangled Regression (MDR) to explore these multiple relations. Specifically, GAT combines a GCN and a graph-aware self-attention in parallel to capture physical and hidden joint-joint relations. Furthermore, MDR models joint-vertex and vertex-vertex interactions to explore joint and vertex relations. Based on the clustering characteristics of vertex offset fields, MDR regresses the vertices by composing the predicted base motions. Extensive experiments show that GATOR achieves state-of-the-art performance on two challenging benchmarks.

cs.CV↗
arXiv2023

M3AE: Multimodal Representation Learning for Brain Tumor Segmentation with Missing Modalities

Multimodal magnetic resonance imaging (MRI) provides complementary information for sub-region analysis of brain tumors. Plenty of methods have been proposed for automatic brain tumor segmentation using four common MRI modalities and achieved remarkable performance. In practice, however, it is common to have one or more modalities missing due to image corruption, artifacts, acquisition protocols, allergy to contrast agents, or simply cost. In this work, we propose a novel two-stage framework for brain tumor segmentation with missing modalities. In the first stage, a multimodal masked autoencoder (M3AE) is proposed, where both random modalities (i.e., modality dropout) and random patches of the remaining modalities are masked for a reconstruction task, for self-supervised learning of robust multimodal representations against missing modalities. To this end, we name our framework M3AE. Meanwhile, we employ model inversion to optimize a representative full-modal image at marginal extra cost, which will be used to substitute for the missing modalities and boost performance during inference. Then in the second stage, a memory-efficient self distillation is proposed to distill knowledge between heterogenous missing-modal situations while fine-tuning the model for supervised segmentation. Our M3AE belongs to the 'catch-all' genre where a single model can be applied to all possible subsets of modalities, thus is economic for both training and deployment. Extensive experiments on BraTS 2018 and 2020 datasets demonstrate its superior performance to existing state-of-the-art methods with missing modalities, as well as the efficacy of its components. Our code is available at: https://github.com/ccarliu/m3ae.

eess.IV↗
arXiv2023

HTNet: Human Topology Aware Network for 3D Human Pose Estimation

3D human pose estimation errors would propagate along the human body topology and accumulate at the end joints of limbs. Inspired by the backtracking mechanism in automatic control systems, we design an Intra-Part Constraint module that utilizes the parent nodes as the reference to build topological constraints for end joints at the part level. Further considering the hierarchy of the human topology, joint-level and body-level dependencies are captured via graph convolutional networks and self-attentions, respectively. Based on these designs, we propose a novel Human Topology aware Network (HTNet), which adopts a channel-split progressive strategy to sequentially learn the structural priors of the human topology from multiple semantic levels: joint, part, and body. Extensive experiments show that the proposed method improves the estimation accuracy by 18.7% on the end joints of limbs and achieves state-of-the-art results on Human3.6M and MPI-INF-3DHP datasets. Code is available at https://github.com/vefalun/HTNet.

cs.CV↗
arXiv2023

A bandwidth theorem for graph transversals

Given a collection $\mathcal{G}=(G_1,\dots, G_h)$ of graphs on the same vertex set $V$ of size $n$, an $h$-edge graph $H$ on the vertex set $V$ is a $\mathcal{G}$-transversal if there exists a bijection $λ: E(H) \rightarrow [h]$ such that $e\in E(G_{λ(e)})$ for each $e\in E(H)$. The conditions on the minimum degree $δ(\mathcal{G})=\min_{i\in[h]}\{ δ(G_i)\}$ for finding a spanning $\mathcal{G}$-transversal isomorphic to a graph $H$ have been actively studied when $H$ is a Hamilton cycle, an $F$-factor, a spanning tree with maximum degree $o(n/\log n)$ and a power of a Hamilton cycle, etc. In this paper, we determined the asymptotically tight threshold on $δ(\mathcal{G})$ for finding a $\mathcal{G}$-transversal isomorphic to $H$ when $H$ is a general $n$-vertex graph with bounded maximum degree and $o(n)$-bandwidth. This provides a transversal generalization of the celebrated Bandwidth theorem by Böttcher, Schacht and Taraz.

math.CO↗
arXiv2023

AIROGS: Artificial Intelligence for RObust Glaucoma Screening Challenge

The early detection of glaucoma is essential in preventing visual impairment. Artificial intelligence (AI) can be used to analyze color fundus photographs (CFPs) in a cost-effective manner, making glaucoma screening more accessible. While AI models for glaucoma screening from CFPs have shown promising results in laboratory settings, their performance decreases significantly in real-world scenarios due to the presence of out-of-distribution and low-quality images. To address this issue, we propose the Artificial Intelligence for Robust Glaucoma Screening (AIROGS) challenge. This challenge includes a large dataset of around 113,000 images from about 60,000 patients and 500 different screening centers, and encourages the development of algorithms that are robust to ungradable and unexpected input data. We evaluated solutions from 14 teams in this paper, and found that the best teams performed similarly to a set of 20 expert ophthalmologists and optometrists. The highest-scoring team achieved an area under the receiver operating characteristic curve of 0.99 (95% CI: 0.98-0.99) for detecting ungradable images on-the-fly. Additionally, many of the algorithms showed robust performance when tested on three other publicly available datasets. These results demonstrate the feasibility of robust AI-enabled glaucoma screening.

eess.IV↗
arXiv2023

Variational Latent-State GPT for Semi-Supervised Task-Oriented Dialog Systems

Recently, two approaches, fine-tuning large pre-trained language models and variational training, have attracted significant interests, separately, for semi-supervised end-to-end task-oriented dialog (TOD) systems. In this paper, we propose Variational Latent-State GPT model (VLS-GPT), which is the first to combine the strengths of the two approaches. Among many options of models, we propose the generative model and the inference model for variational learning of the end-to-end TOD system, both as auto-regressive language models based on GPT-2, which can be further trained over a mix of labeled and unlabeled dialog data in a semi-supervised manner. Variational training of VLS-GPT is both statistically and computationally more challenging than previous variational learning works for sequential latent variable models, which use turn-level first-order Markovian. The inference model in VLS-GPT is non-Markovian due to the use of the Transformer architecture. In this work, we establish Recursive Monte Carlo Approximation (RMCA) to the variational objective with non-Markovian inference model and prove its unbiasedness. Further, we develop the computational strategy of sampling-then-forward-computation to realize RMCA, which successfully overcomes the memory explosion issue of using GPT in variational learning and speeds up training. Semi-supervised TOD experiments are conducted on two benchmark multi-domain datasets of different languages - MultiWOZ2.1 and CrossWOZ. VLS-GPT is shown to significantly outperform both supervised-only and semi-supervised self-training baselines.

cs.CL↗
arXiv2023

Achieving Domain Generalization in Underwater Object Detection by Domain Mixup and Contrastive Learning

The performance of existing underwater object detection methods degrades seriously when facing domain shift caused by complicated underwater environments. Due to the limitation of the number of domains in the dataset, deep detectors easily memorize a few seen domains, which leads to low generalization ability. There are two common ideas to improve the domain generalization performance. First, it can be inferred that the detector trained on as many domains as possible is domain-invariant. Second, for the images with the same semantic content in different domains, their hidden features should be equivalent. This paper further excavates these two ideas and proposes a domain generalization framework (named DMC) that learns how to generalize across domains from Domain Mixup and Contrastive Learning. First, based on the formation of underwater images, an image in an underwater environment is the linear transformation of another underwater environment. Thus, a style transfer model, which outputs a linear transformation matrix instead of the whole image, is proposed to transform images from one source domain to another, enriching the domain diversity of the training data. Second, mixup operation interpolates different domains on the feature level, sampling new domains on the domain manifold. Third, contrastive loss is selectively applied to features from different domains to force the model to learn domain invariant features but retain the discriminative capacity. With our method, detectors will be robust to domain shift. Also, a domain generalization benchmark S-UODAC2020 for detection is set up to measure the performance of our method. Comprehensive experiments on S-UODAC2020 and two object recognition benchmarks (PACS and VLCS) demonstrate that the proposed method is able to learn domain-invariant representations, and outperforms other domain generalization methods.

cs.CV↗
arXiv2023

Many Hamiltonian subsets in large graphs with given density

A set of vertices in a graph is a Hamiltonian subset if it induces a subgraph containing a Hamiltonian cycle. Kim, Liu, Sharifzadeh and Staden proved that among all graphs with minimum degree $d$, $K_{d+1}$ minimises the number of Hamiltonian subsets. We prove a near optimal lower bound that takes also the order and the structure of a graph into account. For many natural graph classes, it provides a much better bound than the extremal one ($\approx 2^{d+1}$). Among others, our bound implies that an $n$-vertex $C_4$-free graphs with minimum degree $d$ contains at least $n2^{d^{2-o(1)}}$ Hamiltonian subsets.

math.CO↗
arXiv2023

A systematic formulation of chiral anomalous magnetohydrodynamics

We present a new way of deriving effective theories of dynamical electromagnetic fields in general media. It can be used to give a systematic formulation of magnetohydrodynamics (MHD) with strong magnetic fields, including systems with chiral matter and Adler-Bell-Jackiw (ABJ) anomaly. We work in the regime in which velocity and temperature fluctuations can be neglected. The resulting chiral anomalous MHD incorporates and generalizes the chiral magnetic effect, the chiral separation effect, the chiral electric separation effect, as well as recently derived strong-field MHD, all in a single coherent framework. At linearized level, the theory predicts that the chiral magnetic wave does not survive dynamical electromagnetic fields. A different chiral wave, to which we refer as the chiral magnetic electric separation wave, emerges as a result of dynamical versions of the chiral electric separation effect and the chiral magnetic effect. We predict its wave velocity. We also introduce a simple, but solvable nonlinear model to explore the fate of the chiral instability.

hep-ph↗
arXiv2023

Polynomial Schur's theorem

We resolve the Ramsey problem for $\{x,y,z:x+y=p(z)\}$ for all polynomials $p$ over $\mathbb{Z}$. In particular, we characterise all polynomials that are $2$-Ramsey, that is, those $p(z)$ such that any $2$-colouring of $\mathbb{N}$ contains infinitely many monochromatic solutions for $x+y=p(z)$. For polynomials that are not $2$-Ramsey, we characterise all $2$-colourings of $\mathbb{N}$ that are not $2$-Ramsey, revealing that certain divisibility barrier is the only obstruction to $2$-Ramseyness for $x+y=p(z)$.

math.NT↗
arXiv2022

Coherent backscattering in the topological Hall effect

The mutual interplay between electron transport and magnetism has attracted considerable attention in recent years, primarily motivated by strategies to manipulate magnetic degrees of freedom electrically, such as spin-orbit torques and domain wall motion. Within this field the topological Hall effect, which originates from scalar spin chirality, is an example of inter-band quantum coherence induced by real-space inhomogeneous magnetic textures, and its magnitude depends on the winding number and chiral spin features that establish the total topological charge of the system. Remarkably, in the two decades since its discovery, there has been no research on the quantum correction to the topological Hall effect. Here we will show that, unlike the ordinary Hall effect, the inhomogeneous magnetization arising from the spin texture will give additional scattering terms in the kinetic equation, which result in a quantum correction to the topological Hall resistivity. We focus on 2D systems, where weak localization is strongest, and determine the complicated gradient corrections to the Cooperon and kinetic equation. Whereas the weak localization correction to the topological Hall effect is not large in currently known materials, we show that it is experimentally observable in dilute magnetic semiconductors. Our theoretical results will stimulate experiments on the topological Hall effect and fill the theoretical knowledge gap on weak localization corrections to transverse transport.

cond-mat.mes-hall↗
arXiv2022

An Unpaired Cross-modality Segmentation Framework Using Data Augmentation and Hybrid Convolutional Networks for Segmenting Vestibular Schwannoma and Cochlea

The crossMoDA challenge aims to automatically segment the vestibular schwannoma (VS) tumor and cochlea regions of unlabeled high-resolution T2 scans by leveraging labeled contrast-enhanced T1 scans. The 2022 edition extends the segmentation task by including multi-institutional scans. In this work, we proposed an unpaired cross-modality segmentation framework using data augmentation and hybrid convolutional networks. Considering heterogeneous distributions and various image sizes for multi-institutional scans, we apply the min-max normalization for scaling the intensities of all scans between -1 and 1, and use the voxel size resampling and center cropping to obtain fixed-size sub-volumes for training. We adopt two data augmentation methods for effectively learning the semantic information and generating realistic target domain scans: generative and online data augmentation. For generative data augmentation, we use CUT and CycleGAN to generate two groups of realistic T2 volumes with different details and appearances for supervised segmentation training. For online data augmentation, we design a random tumor signal reducing method for simulating the heterogeneity of VS tumor signals. Furthermore, we utilize an advanced hybrid convolutional network with multi-dimensional convolutions to adaptively learn sparse inter-slice information and dense intra-slice information for accurate volumetric segmentation of VS tumor and cochlea regions in anisotropic scans. On the crossMoDA2022 validation dataset, our method produces promising results and achieves the mean DSC values of 72.47% and 76.48% and ASSD values of 3.42 mm and 0.53 mm for VS tumor and cochlea regions, respectively.

eess.IV↗
arXiv2022

Self-distillation with Online Diffusion on Batch Manifolds Improves Deep Metric Learning

Recent deep metric learning (DML) methods typically leverage solely class labels to keep positive samples far away from negative ones. However, this type of method normally ignores the crucial knowledge hidden in the data (e.g., intra-class information variation), which is harmful to the generalization of the trained model. To alleviate this problem, in this paper we propose Online Batch Diffusion-based Self-Distillation (OBD-SD) for DML. Specifically, we first propose a simple but effective Progressive Self-Distillation (PSD), which distills the knowledge progressively from the model itself during training. The soft distance targets achieved by PSD can present richer relational information among samples, which is beneficial for the diversity of embedding representations. Then, we extend PSD with an Online Batch Diffusion Process (OBDP), which is to capture the local geometric structure of manifolds in each batch, so that it can reveal the intrinsic relationships among samples in the batch and produce better soft distance targets. Note that our OBDP is able to restore the insufficient manifold relationships obtained by the original PSD and achieve significant performance improvement. Our OBD-SD is a flexible framework that can be integrated into state-of-the-art (SOTA) DML methods. Extensive experiments on various benchmarks, namely CUB200, CARS196, and Stanford Online Products, demonstrate that our OBD-SD consistently improves the performance of the existing DML methods on multiple datasets with negligible additional training time, achieving very competitive results. Code: \url{https://github.com/ZelongZeng/OBD-SD_Pytorch}

cs.CV↗
arXiv2022

Ramsey numbers of cycles versus general graphs

The Ramsey number $R(F,H)$ is the minimum number $N$ such that any $N$-vertex graph either contains a copy of $F$ or its complement contains $H$. Burr in 1981 proved a pleasingly general result that for any graph $H$, provided $n$ is sufficiently large, a natural lower bound construction gives the correct Ramsey number involving cycles: $R(C_n,H)=(n-1)(χ(H)-1)+σ(H)$, where $σ(H)$ is the minimum possible size of a colour class in a $χ(H)$-colouring of $H$. Allen, Brightwell and Skokan conjectured that the same should be true already when $n\geq |H|χ(H)$. We improve this 40-year-old result of Burr by giving quantitative bounds of the form $n\geq C|H|\log^4χ(H)$, which is optimal up to the logarithmic factor. In particular, this proves a strengthening of the Allen-Brightwell-Skokan conjecture for all graphs $H$ with large chromatic number.

math.CO↗
arXiv2022

Rainbow cycles in properly edge-colored graphs

We prove that every properly edge-colored $n$-vertex graph with average degree at least $100(\log n)^2$ contains a rainbow cycle, improving upon $(\log n)^{2+o(1)}$ bound due to Tomon. We also prove that every properly colored $n$-vertex graph with at least $10^5 k^2 n^{1+1/k}$ edges contains a rainbow $2k$-cycle, which improves the previous bound $2^{ck^2}n^{1+1/k}$ obtained by Janzer. Our method using homomorphism inequalities and a lopsided regularization lemma also provides a simple way to prove the Erdős--Simonovits supersaturation theorem for even cycles, which may be of independent interest.

math.CO↗
arXiv2022

Same Pre-training Loss, Better Downstream: Implicit Bias Matters for Language Models

Language modeling on large-scale datasets leads to impressive performance gains on various downstream language tasks. The validation pre-training loss (or perplexity in autoregressive language modeling) is often used as the evaluation metric when developing language models since the pre-training loss tends to be well-correlated with downstream performance (which is itself difficult to evaluate comprehensively). Contrary to this conventional wisdom, this paper shows that 1) pre-training loss cannot fully explain downstream performance and 2) flatness of the model is well-correlated with downstream performance where pre-training loss is not. On simplified datasets, we identify three ways to produce models with the same (statistically optimal) pre-training loss but different downstream performance: continue pre-training after convergence, increasing the model size, and changing the training algorithm. These experiments demonstrate the existence of implicit bias of pre-training algorithms/optimizers -- among models with the same minimal pre-training loss, they implicitly prefer more transferable ones. Toward understanding this implicit bias, we prove that SGD with standard mini-batch noise implicitly prefers flatter minima in language models, and empirically observe a strong correlation between flatness and downstream performance among models with the same minimal pre-training loss. We also prove in a synthetic language setting that among the models with the minimal pre-training loss, the flattest model transfers to downstream tasks.

cs.LG↗
arXiv2022

Information Extraction and Human-Robot Dialogue towards Real-life Tasks: A Baseline Study with the MobileCS Dataset

Recently, there have merged a class of task-oriented dialogue (TOD) datasets collected through Wizard-of-Oz simulated games. However, the Wizard-of-Oz data are in fact simulated data and thus are fundamentally different from real-life conversations, which are more noisy and casual. Recently, the SereTOD challenge is organized and releases the MobileCS dataset, which consists of real-world dialog transcripts between real users and customer-service staffs from China Mobile. Based on the MobileCS dataset, the SereTOD challenge has two tasks, not only evaluating the construction of the dialogue system itself, but also examining information extraction from dialog transcripts, which is crucial for building the knowledge base for TOD. This paper mainly presents a baseline study of the two tasks with the MobileCS dataset. We introduce how the two baselines are constructed, the problems encountered, and the results. We anticipate that the baselines can facilitate exciting future research to build human-robot dialogue systems for real-life tasks.

cs.CL↗