arXiv subjects

Yu Tian

Publications and source records attributed to Yu Tian.

At least 235 records · Page 13Linked to original sources

arXiv2023

Critical scalarization and descalarization of black holes in a generalized scalar-tensor theory

We study the critical dynamics in the scalarization and descalarization in the fully nonlinear dynamical evolution in a general theory with scalar field coupling with both Gauss-Bonnet invariant and Ricci scalar. We explore how the Gauss-Bonnet term triggers the black hole scalarization. A typical type I critical phenomenon is observed, where an unstable critical solution emerges at the threshold and acts as an attractor in the dynamical scalarization. In the descalarization, a marginally stable attractor exists at the threshold of the first order phase transition in shedding off black hole hair. This is a new type I critical phenomenon in the black hole phase transition. Implications of these findings are discussed from the perspective of thermodynamic properties and perturbations for static solutions. We examine the effect of scalar-Ricci coupling on the hyperbolicity in the fully nonlinear evolution and find that such coupling can suppress the elliptic region and enlarge parameter space in computations.

gr-qc↗

arXiv2023

Learning Support and Trivial Prototypes for Interpretable Image Classification

Prototypical part network (ProtoPNet) methods have been designed to achieve interpretable classification by associating predictions with a set of training prototypes, which we refer to as trivial prototypes because they are trained to lie far from the classification boundary in the feature space. Note that it is possible to make an analogy between ProtoPNet and support vector machine (SVM) given that the classification from both methods relies on computing similarity with a set of training points (i.e., trivial prototypes in ProtoPNet, and support vectors in SVM). However, while trivial prototypes are located far from the classification boundary, support vectors are located close to this boundary, and we argue that this discrepancy with the well-established SVM theory can result in ProtoPNet models with inferior classification accuracy. In this paper, we aim to improve the classification of ProtoPNet with a new method to learn support prototypes that lie near the classification boundary in the feature space, as suggested by the SVM theory. In addition, we target the improvement of classification results with a new model, named ST-ProtoPNet, which exploits our support prototypes and the trivial prototypes to provide more effective classification. Experimental results on CUB-200-2011, Stanford Cars, and Stanford Dogs datasets demonstrate that ST-ProtoPNet achieves state-of-the-art classification accuracy and interpretability results. We also show that the proposed support prototypes tend to be better localised in the object of interest rather than in the background region.

cs.CV↗

arXiv2023

On the flame transfer function models for laminar premixed conical and V- flames considering the stretch effect

This paper investigates a predictive model that considers the impact of stretch on the dynamic response of laminar premixed conical and V- flames; the flame stretch consists of two components: the flame curvature and flow strain. The steady and perturbed flame fronts are determined via the linearized $G$-equation associated with the flame stretch model. Parameter analyses of the effects of Markstein length $\mathcal{L}$, flame radius $R$ and unstretched flame aspect ratio $β$ are also conducted. Results show that the flame stretch reduces the steady flame height, with this effect being more significant for larger Markstein lengths and smaller flame sizes. The effects of flame stretch on perturbed flames are evaluated by comparing the flame transfer function (FTF) considering the flame stretch and not. For flames of different sizes, the impact of flame stretch on FTF gain can be divided into three regions. When both $β$ and $R$ are relatively small, due to the decrease in steady flame height and the impact of flow strain, the FTF gain increases. As $β$ and $R$ gradually increase, the FTF gain of the conical flame oscillates periodically while the FTF gain of the V-flame decreases, primarily due to the flame curvature enhancing the flame front disturbance and the wrinkle counteracting effect. When $β$ and $R$ are large, a disruption in wrinkle counteracting effect ensues, leading to a significant increase in FTF gain. Furthermore, as the actual flame height is reduced, the flame stretch also reduces the FTF phase lag which is related to the disturbance propagation time from the flame root to the tip.

physics.flu-dyn↗

arXiv2023

How Robust is Google's Bard to Adversarial Image Attacks?

Multimodal Large Language Models (MLLMs) that integrate text and other modalities (especially vision) have achieved unprecedented performance in various multimodal tasks. However, due to the unsolved adversarial robustness problem of vision models, MLLMs can have more severe safety and security risks by introducing the vision inputs. In this work, we study the adversarial robustness of Google's Bard, a competitive chatbot to ChatGPT that released its multimodal capability recently, to better understand the vulnerabilities of commercial MLLMs. By attacking white-box surrogate vision encoders or MLLMs, the generated adversarial examples can mislead Bard to output wrong image descriptions with a 22% success rate based solely on the transferability. We show that the adversarial examples can also attack other MLLMs, e.g., a 26% attack success rate against Bing Chat and a 86% attack success rate against ERNIE bot. Moreover, we identify two defense mechanisms of Bard, including face detection and toxicity detection of images. We design corresponding attacks to evade these defenses, demonstrating that the current defenses of Bard are also vulnerable. We hope this work can deepen our understanding on the robustness of MLLMs and facilitate future research on defenses. Our code is available at https://github.com/thu-ml/Attack-Bard. Update: GPT-4V is available at October 2023. We further evaluate its robustness under the same set of adversarial examples, achieving a 45% attack success rate.

cs.CV↗

arXiv2023

Unleashing 3D Connectivity in Beyond 5G Networks with Reconfigurable Intelligent Surfaces

Reconfigurable intelligent surfaces (RISs) bring various benefits to the current and upcoming wireless networks, including enhanced spectrum and energy efficiency, soft handover, transmission reliability, and even localization accuracy. These remarkable improvements result from the reconfigurability, programmability, and adaptation capabilities of RISs for fine-tuning radio propagation environments, which can be realized in a cost- and energy-efficient manner. In this paper, we focus on the upgrade of the existing fifth-generation (5G) cellular network with the introduction of an RIS owning a full-dimensional uniform planar array structure for unleashing advanced three-dimensional connectivity. The deployed RIS is exploited for serving unmanned aerial vehicles (UAVs) flying in the sky with ultra-high data rate, a challenging task to be achieved with conventional base stations (BSs) that are designed mainly to serve ground users. By taking into account the line-of-sight probability for the RIS-UAV and BS-UAV links, we formulate the average achievable rate, analyze the effect of environmental parameters, and make insightful performance comparisons. Simulation results show that the deployment of RISs can bring impressive gains and significantly outperform conventional RIS-free 5G networks.

eess.SP↗

arXiv2023

Just Noticeable Difference Modeling for Face Recognition System

High-quality face images are required to guarantee the stability and reliability of automatic face recognition (FR) systems in surveillance and security scenarios. However, a massive amount of face data is usually compressed before being analyzed due to limitations on transmission or storage. The compressed images may lose the powerful identity information, resulting in the performance degradation of the FR system. Herein, we make the first attempt to study just noticeable difference (JND) for the FR system, which can be defined as the maximum distortion that the FR system cannot notice. More specifically, we establish a JND dataset including 3530 original images and 137,670 compressed images generated by advanced reference encoding/decoding software based on the Versatile Video Coding (VVC) standard (VTM-15.0). Subsequently, we develop a novel JND prediction model to directly infer JND images for the FR system. In particular, in order to maximum redundancy removal without impairment of robust identity information, we apply the encoder with multiple feature extraction and attention-based feature decomposition modules to progressively decompose face features into two uncorrelated components, i.e., identity and residual features, via self-supervised learning. Then, the residual feature is fed into the decoder to generate the residual map. Finally, the predicted JND map is obtained by subtracting the residual map from the original image. Experimental results have demonstrated that the proposed model achieves higher accuracy of JND map prediction compared with the state-of-the-art JND models, and is capable of saving more bits while maintaining the performance of the FR system compared with VTM-15.0.

cs.CV↗

arXiv2023

The Devil is in the Details: A Deep Dive into the Rabbit Hole of Data Filtering

The quality of pre-training data plays a critical role in the performance of foundation models. Popular foundation models often design their own recipe for data filtering, which makes it hard to analyze and compare different data filtering approaches. DataComp is a new benchmark dedicated to evaluating different methods for data filtering. This paper describes our learning and solution when participating in the DataComp challenge. Our filtering strategy includes three stages: single-modality filtering, cross-modality filtering, and data distribution alignment. We integrate existing methods and propose new solutions, such as computing CLIP score on horizontally flipped images to mitigate the interference of scene text, using vision and language models to retrieve training samples for target downstream tasks, rebalancing the data distribution to improve the efficiency of allocating the computational budget, etc. We slice and dice our design choices, provide in-depth analysis, and discuss open questions. Our approach outperforms the best method from the DataComp paper by over 4% on the average performance of 38 tasks and by over 2% on ImageNet.

cs.CV↗

arXiv2023

Multimodal Transformers for Wireless Communications: A Case Study in Beam Prediction

Wireless communications at high-frequency bands with large antenna arrays face challenges in beam management, which can potentially be improved by multimodality sensing information from cameras, LiDAR, radar, and GPS. In this paper, we present a multimodal transformer deep learning framework for sensing-assisted beam prediction. We employ a convolutional neural network to extract the features from a sequence of images, point clouds, and radar raw data sampled over time. At each convolutional layer, we use transformer encoders to learn the hidden relations between feature tokens from different modalities and time instances over abstraction space and produce encoded vectors for the next-level feature extraction. We train the model on a combination of different modalities with supervised learning. We try to enhance the model over imbalanced data by utilizing focal loss and exponential moving average. We also evaluate data processing and augmentation techniques such as image enhancement, segmentation, background filtering, multimodal data flipping, radar signal transformation, and GPS angle calibration. Experimental results show that our solution trained on image and GPS data produces the best distance-based accuracy of predicted beams at 78.44%, with effective generalization to unseen day scenarios near 73% and night scenarios over 84%. This outperforms using other modalities and arbitrary data processing techniques, which demonstrates the effectiveness of transformers with feature fusion in performing radio beam prediction from images and GPS. Furthermore, our solution could be pretrained from large sequences of multimodality wireless data, on fine-tuning for multiple downstream radio network tasks.

eess.SP↗

arXiv2023

Free energy, stability, and particle source in dynamical holography

By using the conserved currents in the bulk space-time, we study dynamical holographic systems and the relation between thermodynamical and dynamical stability of such systems. In particular, in the probe limit a generalized free energy is defined with the property of monotonic decreasing in dynamic processes. It is then shown that the (absolute) thermodynamical stability implies the dynamical stability, while the linear dynamical stability implies the thermodynamical (meta-)stability. The holographic superfluid is taken as an example to illustrate our general formalism, where the dynamic evolution of the system in contact with a particle source is clarified by theoretical investigation and numerical verification. The case going beyond the probe limit is also discussed.

hep-th↗

arXiv2023

Harvard Glaucoma Detection and Progression: A Multimodal Multitask Dataset and Generalization-Reinforced Semi-Supervised Learning

Glaucoma is the number one cause of irreversible blindness globally. A major challenge for accurate glaucoma detection and progression forecasting is the bottleneck of limited labeled patients with the state-of-the-art (SOTA) 3D retinal imaging data of optical coherence tomography (OCT). To address the data scarcity issue, this paper proposes two solutions. First, we develop a novel generalization-reinforced semi-supervised learning (SSL) model called pseudo supervisor to optimally utilize unlabeled data. Compared with SOTA models, the proposed pseudo supervisor optimizes the policy of predicting pseudo labels with unlabeled samples to improve empirical generalization. Our pseudo supervisor model is evaluated with two clinical tasks consisting of glaucoma detection and progression forecasting. The progression forecasting task is evaluated both unimodally and multimodally. Our pseudo supervisor model demonstrates superior performance than SOTA SSL comparison models. Moreover, our model also achieves the best results on the publicly available LAG fundus dataset. Second, we introduce the Harvard Glaucoma Detection and Progression (Harvard-GDP) Dataset, a multimodal multitask dataset that includes data from 1,000 patients with OCT imaging data, as well as labels for glaucoma detection and progression. This is the largest glaucoma detection dataset with 3D OCT imaging data and the first glaucoma progression forecasting dataset that is publicly available. Detailed sex and racial analysis are provided, which can be used by interested researchers for fairness learning studies. Our released dataset is benchmarked with several SOTA supervised CNN and transformer deep learning models. The dataset and code are made publicly available via \url{https://ophai.hms.harvard.edu/datasets/harvard-gdp1000}.

cs.CV↗

arXiv2023

Unsupervised Anomaly Detection in Medical Images with a Memory-augmented Multi-level Cross-attentional Masked Autoencoder

Unsupervised anomaly detection (UAD) aims to find anomalous images by optimising a detector using a training set that contains only normal images. UAD approaches can be based on reconstruction methods, self-supervised approaches, and Imagenet pre-trained models. Reconstruction methods, which detect anomalies from image reconstruction errors, are advantageous because they do not rely on the design of problem-specific pretext tasks needed by self-supervised approaches, and on the unreliable translation of models pre-trained from non-medical datasets. However, reconstruction methods may fail because they can have low reconstruction errors even for anomalous images. In this paper, we introduce a new reconstruction-based UAD approach that addresses this low-reconstruction error issue for anomalous images. Our UAD approach, the memory-augmented multi-level cross-attentional masked autoencoder (MemMC-MAE), is a transformer-based approach, consisting of a novel memory-augmented self-attention operator for the encoder and a new multi-level cross-attention operator for the decoder. MemMCMAE masks large parts of the input image during its reconstruction, reducing the risk that it will produce low reconstruction errors because anomalies are likely to be masked and cannot be reconstructed. However, when the anomaly is not masked, then the normal patterns stored in the encoder's memory combined with the decoder's multi-level cross attention will constrain the accurate reconstruction of the anomaly. We show that our method achieves SOTA anomaly detection and localisation on colonoscopy, pneumonia, and covid-19 chest x-ray datasets.

eess.IV↗

arXiv2023

Residual Pattern Learning for Pixel-wise Out-of-Distribution Detection in Semantic Segmentation

Semantic segmentation models classify pixels into a set of known (``in-distribution'') visual classes. When deployed in an open world, the reliability of these models depends on their ability not only to classify in-distribution pixels but also to detect out-of-distribution (OoD) pixels. Historically, the poor OoD detection performance of these models has motivated the design of methods based on model re-training using synthetic training images that include OoD visual objects. Although successful, these re-trained methods have two issues: 1) their in-distribution segmentation accuracy may drop during re-training, and 2) their OoD detection accuracy does not generalise well to new contexts (e.g., country surroundings) outside the training set (e.g., city surroundings). In this paper, we mitigate these issues with: (i) a new residual pattern learning (RPL) module that assists the segmentation model to detect OoD pixels without affecting the inlier segmentation performance; and (ii) a novel context-robust contrastive learning (CoroCL) that enforces RPL to robustly detect OoD pixels among various contexts. Our approach improves by around 10\% FPR and 7\% AuPRC the previous state-of-the-art in Fishyscapes, Segment-Me-If-You-Can, and RoadAnomaly datasets. Our code is available at: https://github.com/yyliu01/RPL.

cs.CV↗

arXiv2023

Time evolution of Einstein-Maxwell-scalar black holes after a thermal quench

We employ the holographic quench technique to drive Einstein-Maxwell-scalar (EMs) black holes out of equilibrium and study the real-time dynamics therein. From the fully nonlinear dynamical simulations, a dynamically unstable Reissner-Nordstr$\ddot{\text{o}}$m anti-de Sitter (RN-AdS) black hole can be scalarized spontaneously after an arbitrarily small quench. On the other hand, a dynamically stable scalarized black hole can be descalarized after a quench of sufficient strength. Interestingly, on the way to descalarization, the scalarized black hole behaves like a holographic superfluid, undergoing a dynamical transition from oscillatory to non-oscillatory decay. Such behaviors are related to the spectrums of quasi-normal modes of scalarized black holes, where the dominant mode migrates toward the imaginary axis with increasing quench strength. In addition, due to the $\mathbb Z_{2}$-symmetry preserved by the model, the ground state is degenerate. We find that there exists a threshold for the quench strength that induces a dynamical transition of the gravitational system from one degenerate ground state to the other. Near the threshold, the gravitational system is attracted to an excited state, that is, a RN-AdS black hole with dynamical instability.

gr-qc↗

arXiv2023

Self-supervised Pseudo Multi-class Pre-training for Unsupervised Anomaly Detection and Segmentation in Medical Images

Unsupervised anomaly detection (UAD) methods are trained with normal (or healthy) images only, but during testing, they are able to classify normal and abnormal (or disease) images. UAD is an important medical image analysis (MIA) method to be applied in disease screening problems because the training sets available for those problems usually contain only normal images. However, the exclusive reliance on normal images may result in the learning of ineffective low-dimensional image representations that are not sensitive enough to detect and segment unseen abnormal lesions of varying size, appearance, and shape. Pre-training UAD methods with self-supervised learning, based on computer vision techniques, can mitigate this challenge, but they are sub-optimal because they do not explore domain knowledge for designing the pretext tasks, and their contrastive learning losses do not try to cluster the normal training images, which may result in a sparse distribution of normal images that is ineffective for anomaly detection. In this paper, we propose a new self-supervised pre-training method for MIA UAD applications, named Pseudo Multi-class Strong Augmentation via Contrastive Learning (PMSACL). PMSACL consists of a novel optimisation method that contrasts a normal image class from multiple pseudo classes of synthesised abnormal images, with each class enforced to form a dense cluster in the feature space. In the experiments, we show that our PMSACL pre-training improves the accuracy of SOTA UAD methods on many MIA benchmarks using colonoscopy, fundus screening and Covid-19 Chest X-ray datasets. The code is made publicly available via https://github.com/tianyu0207/PMSACL.

eess.IV↗

arXiv2023

BoMD: Bag of Multi-label Descriptors for Noisy Chest X-ray Classification

Deep learning methods have shown outstanding classification accuracy in medical imaging problems, which is largely attributed to the availability of large-scale datasets manually annotated with clean labels. However, given the high cost of such manual annotation, new medical imaging classification problems may need to rely on machine-generated noisy labels extracted from radiology reports. Indeed, many Chest X-ray (CXR) classifiers have already been modelled from datasets with noisy labels, but their training procedure is in general not robust to noisy-label samples, leading to sub-optimal models. Furthermore, CXR datasets are mostly multi-label, so current noisy-label learning methods designed for multi-class problems cannot be easily adapted. In this paper, we propose a new method designed for the noisy multi-label CXR learning, which detects and smoothly re-labels samples from the dataset, which is then used to train common multi-label classifiers. The proposed method optimises a bag of multi-label descriptors (BoMD) to promote their similarity with the semantic descriptors produced by BERT models from the multi-label image annotation. Our experiments on diverse noisy multi-label training sets and clean testing sets show that our model has state-of-the-art accuracy and robustness in many CXR multi-label classification benchmarks.

eess.IV↗

arXiv2023

Baby Skyrmion in two-component holographic superfluids

In the two-component Ginzburg-Landau theory of superfluidity, a pair of fractional vortices form a composite type of topological defect, usually referred to as a baby skyrmion. In this paper, we initiate the construction of such a baby skyrmion in the holographic model of two-component superfluids. As a result, two types of baby skyrmion configurations are found, where the monopole-type of one is constructed directly by solving the static equations of motion while the dipole-type of one is obtained by resorting to the time evolution method. In addition, we find that the existence of these two types of baby skyrmion depends on the inter-component coupling, reminiscent of the situation in the baby skyrmion model.

hep-th↗

arXiv2023

Why Is Prompt Tuning for Vision-Language Models Robust to Noisy Labels?

Vision-language models such as CLIP learn a generic text-image embedding from large-scale training data. A vision-language model can be adapted to a new classification task through few-shot prompt tuning. We find that such a prompt tuning process is highly robust to label noises. This intrigues us to study the key reasons contributing to the robustness of the prompt tuning paradigm. We conducted extensive experiments to explore this property and find the key factors are: 1) the fixed classname tokens provide a strong regularization to the optimization of the model, reducing gradients induced by the noisy samples; 2) the powerful pre-trained image-text embedding that is learned from diverse and generic web data provides strong prior knowledge for image classification. Further, we demonstrate that noisy zero-shot predictions from CLIP can be used to tune its own prompt, significantly enhancing prediction accuracy in the unsupervised setting. The code is available at https://github.com/CEWu/PTNL.

cs.CV↗

arXiv2023

Curvature-based Clustering on Graphs

Unsupervised node clustering (or community detection) is a classical graph learning task. In this paper, we study algorithms, which exploit the geometry of the graph to identify densely connected substructures, which form clusters or communities. Our method implements discrete Ricci curvatures and their associated geometric flows, under which the edge weights of the graph evolve to reveal its community structure. We consider several discrete curvature notions and analyze the utility of the resulting algorithms. In contrast to prior literature, we study not only single-membership community detection, where each node belongs to exactly one community, but also mixed-membership community detection, where communities may overlap. For the latter, we argue that it is beneficial to perform community detection on the line graph, i.e., the graph's dual. We provide both theoretical and empirical evidence for the utility of our curvature-based clustering algorithms. In addition, we give several results on the relationship between the curvature of a graph and that of its dual, which enable the efficient implementation of our proposed mixed-membership community detection approach and which may be of independent interest for curvature-based network analysis.

cs.SI↗