arXiv subjects

Ruijie Yang

Publications and source records attributed to Ruijie Yang.

34 records · Page 2Linked to original sources

arXiv2024

Leveraging Predicate and Triplet Learning for Scene Graph Generation

Scene Graph Generation (SGG) aims to identify entities and predict the relationship triplets \textit{\textless subject, predicate, object\textgreater } in visual scenes. Given the prevalence of large visual variations of subject-object pairs even in the same predicate, it can be quite challenging to model and refine predicate representations directly across such pairs, which is however a common strategy adopted by most existing SGG methods. We observe that visual variations within the identical triplet are relatively small and certain relation cues are shared in the same type of triplet, which can potentially facilitate the relation learning in SGG. Moreover, for the long-tail problem widely studied in SGG task, it is also crucial to deal with the limited types and quantity of triplets in tail predicates. Accordingly, in this paper, we propose a Dual-granularity Relation Modeling (DRM) network to leverage fine-grained triplet cues besides the coarse-grained predicate ones. DRM utilizes contexts and semantics of predicate and triplet with Dual-granularity Constraints, generating compact and balanced representations from two perspectives to facilitate relation recognition. Furthermore, a Dual-granularity Knowledge Transfer (DKT) strategy is introduced to transfer variation from head predicates/triplets to tail ones, aiming to enrich the pattern diversity of tail classes to alleviate the long-tail problem. Extensive experiments demonstrate the effectiveness of our method, which establishes new state-of-the-art performance on Visual Genome, Open Image, and GQA datasets. Our code is available at \url{https://github.com/jkli1998/DRM}

cs.CV↗

arXiv2023

Global Contrast Masked Autoencoders Are Powerful Pathological Representation Learners

Based on digital pathology slice scanning technology, artificial intelligence algorithms represented by deep learning have achieved remarkable results in the field of computational pathology. Compared to other medical images, pathology images are more difficult to annotate, and thus, there is an extreme lack of available datasets for conducting supervised learning to train robust deep learning models. In this paper, we propose a self-supervised learning (SSL) model, the global contrast-masked autoencoder (GCMAE), which can train the encoder to have the ability to represent local-global features of pathological images, also significantly improve the performance of transfer learning across data sets. In this study, the ability of the GCMAE to learn migratable representations was demonstrated through extensive experiments using a total of three different disease-specific hematoxylin and eosin (HE)-stained pathology datasets: Camelyon16, NCTCRC and BreakHis. In addition, this study designed an effective automated pathology diagnosis process based on the GCMAE for clinical applications. The source code of this paper is publicly available at https://github.com/StarUniversus/gcmae.

eess.IV↗

arXiv2023

Hirzebruch-Milnor classes of hypersurfaces with nontrivial normal bundles and applications to higher du Bois and rational singularities

We extend the Hirzebruch-Milnor class of a hypersurface $X$ to the case where the normal bundle is nontrivial and $X$ cannot be defined by a global function, using the associated line bundle and the graded quotients of the monodromy filtration. The earlier definition requiring a global defining function of $X$ can be applied rarely to projective hypersurfaces with non-isolated singularities. Indeed, it is surprisingly difficult to get a one-parameter smoothing with total space smooth without destroying the singularities by blowing-ups (except certain quite special cases). As an application, assuming the singular locus is a projective variety, we show that the minimal exponent of a hypersurface can be captured by the spectral Hirzebruch-Milnor class, and higher du~Bois and rational singularities of a hypersurface are detectable by the unnormalized Hirzebruch-Milnor class. Here the unnormalized class can be replaced by the normalized one in the higher du~Bois case, but for the higher rational case, we must use also the decomposition of the Hirzebruch-Milnor class by the action of the semisimple part of the monodromy (which is equivalent to the spectral Hirzebruch-Milnor class). We cannot extend these arguments to the non-projective compact case by Hironaka's example.

math.AG↗

arXiv2023

Cohomology of semisimple local systems and the Decomposition theorem

In this paper, we study the cohomology of semisimple local systems in the spirit of classical Hodge theory. On the one hand, we establish a generalization of Hodge-Riemann bilinear relations. For a semisimple local system on a smooth projective variety, we define a canonical isomorphism from the complex conjugate of its cohomology to the cohomology of the dual local system, which is a generalization of the classical Weil operator for pure Hodge structures. This isomorphism establishes a relation between the twisted Poincaré pairing, a purely topological object, and a positive definite Hermitian pairing. On the other hand, we prove a global invariant cycle theorem for semisimple local systems. As an application, we give a new and geometric proof of Sabbah's Decomposition Theorem for the direct images of semisimple local systems under proper algebraic maps, without using the category of polarizable twistor D-modules.

math.AG↗

arXiv2023

Common Knowledge Learning for Generating Transferable Adversarial Examples

This paper focuses on an important type of black-box attacks, i.e., transfer-based adversarial attacks, where the adversary generates adversarial examples by a substitute (source) model and utilize them to attack an unseen target model, without knowing its information. Existing methods tend to give unsatisfactory adversarial transferability when the source and target models are from different types of DNN architectures (e.g. ResNet-18 and Swin Transformer). In this paper, we observe that the above phenomenon is induced by the output inconsistency problem. To alleviate this problem while effectively utilizing the existing DNN models, we propose a common knowledge learning (CKL) framework to learn better network weights to generate adversarial examples with better transferability, under fixed network architectures. Specifically, to reduce the model-specific features and obtain better output distributions, we construct a multi-teacher framework, where the knowledge is distilled from different teacher architectures into one student network. By considering that the gradient of input is usually utilized to generated adversarial examples, we impose constraints on the gradients between the student and teacher models, to further alleviate the output inconsistency problem and enhance the adversarial transferability. Extensive experiments demonstrate that our proposed work can significantly improve the adversarial transferability.

cs.LG↗

arXiv2023

Higher Du Bois and higher rational singularities of hypersurfaces

In this note, we give several equivalent characterizations of higher Du Bois and higher rational singularities in the context of globally defined hypersurfaces. As a key input, we characterize these singularities using the Hodge filtration on the vanishing cycle mixed Hodge module. As an application, we indicate a homological criterion for detecting higher Du Bois and higher rational hypersurface singularities, formulated in terms of the spectral Hirzebruch-Milnor characteristic classes.

math.AG↗

arXiv2023

Deep Reinforcement Learning for Beam Angle Optimization of Intensity-Modulated Radiation Therapy

Objective: Intensity-modulated radiation therapy (IMRT) beam angle optimization (BAO) is a challenging combinatorial optimization problem that is NP-hard. In this study, we aim to develop a personalized BAO algorithm for IMRT that improves the quality of the final treatment. Methods: To improve the quality of IMRT treatment planning, we propose a deep reinforcement learning (DRL)-based approach for IMRT BAO. We consider the task as a sequential decision-making problem and formulate it as a Markov Decision Process. To facilitate the training process, a 3D-Unet is designed to predict the dose distribution for the different number of beam angles, ranging from 1 to 9, to simulate the IMRT environment. By leveraging the simulation model, double deep-Q network (DDQN) and proximal policy optimization (PPO) are used to train agents to select the personalized beam angle sequentially within a few seconds. Results: The treatment plans with beam angles selected by DRL outperform those with clinically used evenly distributed beam angles. For DDQN, the overall average improvement of the CIs is 0.027, 0.032, and 0.03 for 5, 7, and 9 beam angles respectively. For PPO, the overall average improvement of CIs is 0.045, 0.051, and 0.025 for 5, 7, and 9 beam angles respectively. Conclusion: The proposed DRL-based beam angle selection strategy can generate personalized beam angles within a few seconds, and the resulting treatment plan is superior to that obtained using evenly distributed angles. Significance: A fast and automated personalized beam angle selection approach is been proposed for IMRT BAO.

physics.med-ph↗

arXiv2022

Zeros of one-forms and homologically trivial fibrations

We show that a conjecture of Kotschick about one-forms without zeros on compact Kähler manifolds follows in the case of simple Albanese torus from a conjecture of Bobadilla and Kollár about homologically trivial fibrations. As an application, we prove Kotschick's conjecture for compact Kähler manifolds X whose first betti number is at least 2dim(X)-2 and Albanese torus is simple.

math.AG↗

arXiv2022

Hodge modules and Singular Hermitian Metrics

The purpose of this paper is to study certain notions of metric positivity for the lowest nonzero piece in the Hodge filtration of a Hodge module. We show that the Hodge metric satisfies the minimal extension property. In particular, this singular Hermitian metric has semi-positive curvature.

math.AG↗

arXiv2022

Nodal elliptic curves on K3 surfaces

Let $(X,L)$ be a general primitively polarized K3 surface with $c_1(L)^2 = 2g-2$ for some integer $g \geq 2$. The Severi variety $V^{L,δ} \subset |L|$ is defined to be the locus of reduced and irreducible curves in $|L|$ with exactly $δ$ nodes and no other singularities. When $δ=g$, any curve $C \in V^{L,g}$ is a rational curve; in fact, Chen \cite{Chen02} has shown that all rational curves in $|L|$ are nodal, and the number of such rational curves is given by the Yau-Zaslow formula \cite{YZ96}. In this paper, we consider the next case where $δ= g-1$ and the Severi variety $V^{L,g-1}$ parametrizing nodal elliptic curves is of dimension 1. Let $\overline{V}^{L,g-1} \subset |L|$ denote the Zariski closure. For a reduced curve $C$, we define the geometric genus of $C$ to be the sum of the genera of the irreducible components of the normalization. We prove that the geometric genus of the closure $\overline{V}^{L,g-1} \subset |L|$ is bounded from below by $O(e^{C\sqrt{g}})$.

math.AG↗

arXiv2021

A Perceptual Distortion Reduction Framework: Towards Generating Adversarial Examples with High Perceptual Quality and Attack Success Rate

Most of the adversarial attack methods suffer from large perceptual distortions such as visible artifacts, when the attack strength is relatively high. These perceptual distortions contain a certain portion which contributes less to the attack success rate. This portion of distortions, which is induced by unnecessary modifications and lack of proper perceptual distortion constraint, is the target of the proposed framework. In this paper, we propose a perceptual distortion reduction framework to tackle this problem from two perspectives. Firstly, we propose a perceptual distortion constraint and add it into the objective function to jointly optimize the perceptual distortions and attack success rate. Secondly, we propose an adaptive penalty factor $λ$ to balance the discrepancies between different samples. Since SGD and Momentum-SGD cannot optimize our complex non-convex problem, we exploit Adam in optimization. Extensive experiments have verified the superiority of our proposed framework.

cs.CV↗

arXiv2021

iDARTS: Improving DARTS by Node Normalization and Decorrelation Discretization

Differentiable ARchiTecture Search (DARTS) uses a continuous relaxation of network representation and dramatically accelerates Neural Architecture Search (NAS) by almost thousands of times in GPU-day. However, the searching process of DARTS is unstable, which suffers severe degradation when training epochs become large, thus limiting its application. In this paper, we claim that this degradation issue is caused by the imbalanced norms between different nodes and the highly correlated outputs from various operations. We then propose an improved version of DARTS, namely iDARTS, to deal with the two problems. In the training phase, it introduces node normalization to maintain the norm balance. In the discretization phase, the continuous architecture is approximated based on the similarity between the outputs of the node and the decorrelated operations rather than the values of the architecture parameters. Extensive evaluation is conducted on CIFAR-10 and ImageNet, and the error rates of 2.25\% and 24.7\% are reported within 0.2 and 1.9 GPU-day for architecture search respectively, which shows its effectiveness. Additional analysis also reveals that iDARTS has the advantage in robustness and generalization over other DARTS-based counterparts.

cs.CV↗

arXiv2021

Exploring Transferable and Robust Adversarial Perturbation Generation from the Perspective of Network Hierarchy

The transferability and robustness of adversarial examples are two practical yet important properties for black-box adversarial attacks. In this paper, we explore effective mechanisms to boost both of them from the perspective of network hierarchy, where a typical network can be hierarchically divided into output stage, intermediate stage and input stage. Since over-specialization of source model, we can hardly improve the transferability and robustness of the adversarial perturbations in the output stage. Therefore, we focus on the intermediate and input stages in this paper and propose a transferable and robust adversarial perturbation generation (TRAP) method. Specifically, we propose the dynamically guided mechanism to continuously calculate accurate directional guidances for perturbation generation in the intermediate stage. In the input stage, instead of the single-form transformation augmentations adopted in the existing methods, we leverage multiform affine transformation augmentations to further enrich the input diversity and boost the robustness and transferability of the adversarial perturbations. Extensive experiments demonstrate that our TRAP achieves impressive transferability and high robustness against certain interferences.

cs.CV↗

arXiv2020

On irrationality of hypersurfaces In $\mathbf{P}^{n+1}$

In this note, we study various measures of irrationality for hypersurfaces in projective spaces which were recently proposed by Bastianelli, De Poi, Ein, Lazarsfeld and Ullery. In particular, we answer the question raised by Bastianelli that if $X \subset P^{n+1}$ is a very general smooth hypersurface of dimension $n$ and degree $d\geq 2n+2$, then $\text{stab.irr}(X)=\text{uni.irr}(X)=d-1$. As a corollary, we prove that $\text{irr}(X\times P^{m})=\text{irr}(X)$ for any integer $m\geq 1$.

math.AG↗

arXiv2020

$L^2$-minimal extensions over Hermitian symmetric domains

In this paper, we study the $L^2$-minimal extension problem for polarized variations of Hodge structures over Hermitian symmetric domains. We are able to explicitly find the $L^2$-minimal extensions using a group-theoretic construction. In particular, this gives a construction without using $L^2$-estimates as in the Ohsawa-Takegoshi type extension theorems. The key ingredient is the Harish-Chandra embedding of Hermitian symmetric domains. The construction of holomorphic sections might be of independent interest since it gives a concrete description in the setting of Hermitian VHS.

math.AG↗

arXiv2020

Nef cones of nested Hilbert schemes of points on surfaces

Let $X$ be the projective plane, a Hirzebruch surface, or a general $K3$ surface. In this paper, we study the birational geometry of various nested Hilbert schemes of points parameterizing pairs of zero-dimensional subschemes on $X$. We calculate the nef cone for two types of nested Hilbert schemes: $X^{[n+1,n]}$ and universal families. As an application, we recover a theorem of Butler on syzygies on Hirzebruch surfaces.

math.AG↗