arXiv subjects

Hao Chen

Publications and source records attributed to Hao Chen.

At least 1,063 records · Page 59Linked to original sources

arXiv2023

Learning Robust Representation for Joint Grading of Ophthalmic Diseases via Adaptive Curriculum and Feature Disentanglement

Diabetic retinopathy (DR) and diabetic macular edema (DME) are leading causes of permanent blindness worldwide. Designing an automatic grading system with good generalization ability for DR and DME is vital in clinical practice. However, prior works either grade DR or DME independently, without considering internal correlations between them, or grade them jointly by shared feature representation, yet ignoring potential generalization issues caused by difficult samples and data bias. Aiming to address these problems, we propose a framework for joint grading with the dynamic difficulty-aware weighted loss (DAW) and the dual-stream disentangled learning architecture (DETACH). Inspired by curriculum learning, DAW learns from simple samples to difficult samples dynamically via measuring difficulty adaptively. DETACH separates features of grading tasks to avoid potential emphasis on the bias. With the addition of DAW and DETACH, the model learns robust disentangled feature representations to explore internal correlations between DR and DME and achieve better grading performance. Experiments on three benchmarks show the effectiveness and robustness of our framework under both the intra-dataset and cross-dataset tests.

cs.CV↗

arXiv2023

DoNet: Deep De-overlapping Network for Cytology Instance Segmentation

Cell instance segmentation in cytology images has significant importance for biology analysis and cancer screening, while remains challenging due to 1) the extensive overlapping translucent cell clusters that cause the ambiguous boundaries, and 2) the confusion of mimics and debris as nuclei. In this work, we proposed a De-overlapping Network (DoNet) in a decompose-and-recombined strategy. A Dual-path Region Segmentation Module (DRM) explicitly decomposes the cell clusters into intersection and complement regions, followed by a Semantic Consistency-guided Recombination Module (CRM) for integration. To further introduce the containment relationship of the nucleus in the cytoplasm, we design a Mask-guided Region Proposal Strategy (MRP) that integrates the cell attention maps for inner-cell instance prediction. We validate the proposed approach on ISBI2014 and CPS datasets. Experiments show that our proposed DoNet significantly outperforms other state-of-the-art (SOTA) cell instance segmentation methods. The code is available at https://github.com/DeepDoNet/DoNet.

cs.CV↗

arXiv2023

Covering $b$-Symbol Metric Codes and the Generalized Singleton Bound

Symbol-pair codes were proposed for the application in high density storage systems, where it is not possible to read individual symbols. Yaakobi, Bruck and Siegel proved that the minimum pair-distance of binary linear cyclic codes satisfies $d_2 \geq \lceil 3d_H/2 \rceil$ and introduced $b$-symbol metric codes in 2016. In this paper covering codes in $b$-symbol metrics are considered. Some examples are given to show that the Delsarte bound and the Norse bound for covering codes in the Hamming metric are not true for covering codes in the pair metric. We give the redundancy bound on covering radius of linear codes in the $b$-symbol metric and give some optimal codes attaining this bound. Then we prove that there is no perfect linear symbol-pair code with the minimum pair distance $7$ and there is no perfect $b$-symbol metric code if $b\geq \frac{n+1}{2}$. Moreover a lot of cyclic and algebraic-geometric codes are proved non-perfect in the $b$-symbol metric. The covering radius of the Reed-Solomon code in the $b$-symbol metric is determined. As an application the generalized Singleton bound on the sizes of list-decodable $b$-symbol metric codes is also presented. Then an upper bound on lengths of general MDS symbol-pair codes is proved.

cs.IT↗

arXiv2023

Two-level Graph Network for Few-Shot Class-Incremental Learning

Few-shot class-incremental learning (FSCIL) aims to design machine learning algorithms that can continually learn new concepts from a few data points, without forgetting knowledge of old classes. The difficulty lies in that limited data from new classes not only lead to significant overfitting issues but also exacerbates the notorious catastrophic forgetting problems. However, existing FSCIL methods ignore the semantic relationships between sample-level and class-level. % Using the advantage that graph neural network (GNN) can mine rich information among few samples, In this paper, we designed a two-level graph network for FSCIL named Sample-level and Class-level Graph Neural Network (SCGN). Specifically, a pseudo incremental learning paradigm is designed in SCGN, which synthesizes virtual few-shot tasks as new tasks to optimize SCGN model parameters in advance. Sample-level graph network uses the relationship of a few samples to aggregate similar samples and obtains refined class-level features. Class-level graph network aims to mitigate the semantic conflict between prototype features of new classes and old classes. SCGN builds two-level graph networks to guarantee the latent semantic of each few-shot class can be effectively represented in FSCIL. Experiments on three popular benchmark datasets show that our method significantly outperforms the baselines and sets new state-of-the-art results with remarkable advantages.

cs.CV↗

arXiv2023

Towards Scalable Neural Representation for Diverse Videos

Implicit neural representations (INR) have gained increasing attention in representing 3D scenes and images, and have been recently applied to encode videos (e.g., NeRV, E-NeRV). While achieving promising results, existing INR-based methods are limited to encoding a handful of short videos (e.g., seven 5-second videos in the UVG dataset) with redundant visual content, leading to a model design that fits individual video frames independently and is not efficiently scalable to a large number of diverse videos. This paper focuses on developing neural representations for a more practical setup -- encoding long and/or a large number of videos with diverse visual content. We first show that instead of dividing videos into small subsets and encoding them with separate models, encoding long and diverse videos jointly with a unified model achieves better compression results. Based on this observation, we propose D-NeRV, a novel neural representation framework designed to encode diverse videos by (i) decoupling clip-specific visual content from motion information, (ii) introducing temporal reasoning into the implicit neural network, and (iii) employing the task-oriented flow as intermediate output to reduce spatial redundancies. Our new model largely surpasses NeRV and traditional video compression techniques on UCF101 and UVG datasets on the video compression task. Moreover, when used as an efficient data-loader, D-NeRV achieves 3%-10% higher accuracy than NeRV on action recognition tasks on the UCF101 dataset under the same compression ratios.

cs.CV↗

arXiv2023

Exploring Visual Prompts for Whole Slide Image Classification with Multiple Instance Learning

Multiple instance learning (MIL) has emerged as a popular method for classifying histopathology whole slide images (WSIs). However, existing approaches typically rely on pre-trained models from large natural image datasets, such as ImageNet, to generate instance features, which can be sub-optimal due to the significant differences between natural images and histopathology images that lead to a domain shift. In this paper, we present a novel, simple yet effective method for learning domain-specific knowledge transformation from pre-trained models to histopathology images. Our approach entails using a prompt component to assist the pre-trained model in discerning differences between the pre-trained dataset and the target histopathology dataset, resulting in improved performance of MIL models. We validate our method on two publicly available datasets, Camelyon16 and TCGA-NSCLC. Extensive experimental results demonstrate the significant performance improvement of our method for different MIL models and backbones. Upon publication of this paper, we will release the source code for our method.

cs.CV↗

arXiv2023

Dual-distribution discrepancy with self-supervised refinement for anomaly detection in medical images

Medical anomaly detection is a crucial yet challenging task aimed at recognizing abnormal images to assist in diagnosis. Due to the high-cost annotations of abnormal images, most methods utilize only known normal images during training and identify samples deviating from the normal profile as anomalies in the testing phase. Many readily available unlabeled images containing anomalies are thus ignored in the training phase, restricting the performance. To solve this problem, we introduce one-class semi-supervised learning (OC-SSL) to utilize known normal and unlabeled images for training, and propose Dual-distribution Discrepancy for Anomaly Detection (DDAD) based on this setting. Ensembles of reconstruction networks are designed to model the distribution of normal images and the distribution of both normal and unlabeled images, deriving the normative distribution module (NDM) and unknown distribution module (UDM). Subsequently, the intra-discrepancy of NDM and inter-discrepancy between the two modules are designed as anomaly scores. Furthermore, we propose a new perspective on self-supervised learning, which is designed to refine the anomaly scores rather than detect anomalies directly. Five medical datasets, including chest X-rays, brain MRIs and retinal fundus images, are organized as benchmarks for evaluation. Experiments on these benchmarks comprehensively compare a wide range of anomaly detection methods and demonstrate that our method achieves significant gains and outperforms the state-of-the-art. Code and organized benchmarks are available at https://github.com/caiyu6666/DDAD-ASR.

cs.CV↗

arXiv2023

MIGHTEE-\HI: Possible interactions with the galaxy NGC~895

The transformation and evolution of a galaxy is strongly influenced by interactions with its environment. Neutral hydrogen (\HI) is an excellent way to trace these interactions. Here, we present \HI\ observations of the spiral galaxy NGC~895, which was previously thought to be isolated. High-sensitivity \HI\ observations from the MeerKAT large survey project MIGHTEE reveal possible interaction features, such as extended spiral arms, and the two newly discovered \HI\ companions, that drive us to change the narrative that it is an isolated galaxy. We combine these observations with deep optical images from the Hyper Suprime Camera to show an absence of tidal debris between NGC 895 and its companions. We do find an excess of light in the outer parts of the companion galaxy MGTH$\_$J022138.1-052631 which could be an indication of external perturbation and thus possible sign of interactions. Our analysis shows that NGC~895 is an actively star-forming galaxy with a SFR of $\mathrm{1.75 \pm 0.09 [M_{\odot}/yr]}$, a value typical for high stellar mass galaxies on the star forming main sequence. It is reasonable to state that different mechanisms may have contributed to the observed features in NGC~895 and this emphasizes the need to revisit the target with more detailed observations. Our work shows the high potential and synergy of using state-of-the-art data in both \HI\ and optical to reveal a more complete picture of galaxy environments.

astro-ph.GA↗

arXiv2023

SoftMatch: Addressing the Quantity-Quality Trade-off in Semi-supervised Learning

The critical challenge of Semi-Supervised Learning (SSL) is how to effectively leverage the limited labeled data and massive unlabeled data to improve the model's generalization performance. In this paper, we first revisit the popular pseudo-labeling methods via a unified sample weighting formulation and demonstrate the inherent quantity-quality trade-off problem of pseudo-labeling with thresholding, which may prohibit learning. To this end, we propose SoftMatch to overcome the trade-off by maintaining both high quantity and high quality of pseudo-labels during training, effectively exploiting the unlabeled data. We derive a truncated Gaussian function to weight samples based on their confidence, which can be viewed as a soft version of the confidence threshold. We further enhance the utilization of weakly-learned classes by proposing a uniform alignment approach. In experiments, SoftMatch shows substantial improvements across a wide variety of benchmarks, including image, text, and imbalanced classification.

cs.LG↗

arXiv2023

Implicit Ray-Transformers for Multi-view Remote Sensing Image Segmentation

The mainstream CNN-based remote sensing (RS) image semantic segmentation approaches typically rely on massive labeled training data. Such a paradigm struggles with the problem of RS multi-view scene segmentation with limited labeled views due to the lack of considering 3D information within the scene. In this paper, we propose ''Implicit Ray-Transformer (IRT)'' based on Implicit Neural Representation (INR), for RS scene semantic segmentation with sparse labels (such as 4-6 labels per 100 images). We explore a new way of introducing multi-view 3D structure priors to the task for accurate and view-consistent semantic segmentation. The proposed method includes a two-stage learning process. In the first stage, we optimize a neural field to encode the color and 3D structure of the remote sensing scene based on multi-view images. In the second stage, we design a Ray Transformer to leverage the relations between the neural field 3D features and 2D texture features for learning better semantic representations. Different from previous methods that only consider 3D prior or 2D features, we incorporate additional 2D texture information and 3D prior by broadcasting CNN features to different point features along the sampled ray. To verify the effectiveness of the proposed method, we construct a challenging dataset containing six synthetic sub-datasets collected from the Carla platform and three real sub-datasets from Google Maps. Experiments show that the proposed method outperforms the CNN-based methods and the state-of-the-art INR-based segmentation methods in quantitative and qualitative metrics.

cs.CV↗

arXiv2023

A Monkey Swing Counting Algorithm Based on Object Detection

This paper focuses on proposing a deep learning-based monkey swing counting algorithm. Nowadays, there are very few papers on monkey detection, and even fewer papers on monkey swing counting. This research focuses on this gap and attempts to count the number of monkeys swinging their heads by deep learning. This paper further extends the traditional target detection algorithm. By analyzing the results of object detection, we localize the monkey's actions over a period of time. This paper analyzes the task of counting monkey head swings, and proposes the standard that accurately describes a monkey swinging its head. Under the guidance of this standard, the head-swing count in 50 monkey movement videos in this paper has achieved 94%.

cs.CV↗

arXiv2023

Traj-MAE: Masked Autoencoders for Trajectory Prediction

Trajectory prediction has been a crucial task in building a reliable autonomous driving system by anticipating possible dangers. One key issue is to generate consistent trajectory predictions without colliding. To overcome the challenge, we propose an efficient masked autoencoder for trajectory prediction (Traj-MAE) that better represents the complicated behaviors of agents in the driving environment. Specifically, our Traj-MAE employs diverse masking strategies to pre-train the trajectory encoder and map encoder, allowing for the capture of social and temporal information among agents while leveraging the effect of environment from multiple granularities. To address the catastrophic forgetting problem that arises when pre-training the network with multiple masking strategies, we introduce a continual pre-training framework, which can help Traj-MAE learn valuable and diverse information from various strategies efficiently. Our experimental results in both multi-agent and single-agent settings demonstrate that Traj-MAE achieves competitive results with state-of-the-art methods and significantly outperforms our baseline model.

cs.CV↗

arXiv2023

Parallel Computing Based Solution for Reliability-Constrained Distribution Network Planning

The main goal of distribution network (DN) expansion planning is essentially to achieve minimal investment constrained with specified reliability requirements. The reliability-constrained distribution network planning (RcDNP) problem can be cast an instance of mixed-integer linear programming (MILP) which involves ultra-heavy computation burden especially for large scale DNs. In this paper, we propose a parallel computing-based acceleration algorithm for solve RcDNP problem. The RcDNP is decomposed into a backbone grid and several lateral grid problems with coordination. Then a parallelizable augmented Lagrangian algorithm with acceleration strategy is developed to solve the coordination planning problems. In this method, the lateral grid problems are solved in parallel through coordinating with the backbone grid planning problem. To address the presence of nonconvexity, Gauss-Seidel iteration is adopted on the convex hull of the feasible region constructed by the decomposition method. Under mild conditions, the optimality and convergence of this algorithm is proven. The numerical tests show the proposed method can significantly reduce the computation time and make the RcDNP applicable for real-world problems.

eess.SY↗

arXiv2023

On the tightness of graph-based statistics

We establish tightness of graph-based stochastic processes in the space $D[0+ε,1-ε]$ with $ε>0$ that allows for discontinuities of the first kind. The graph-based stochastic processes are based on statistics constructed from similarity graphs. In this setting, the classic characterization of tightness is intractable, making it difficult to obtain convergence of the limiting distributions for graph-based stochastic processes. We take an alternative approach and study the behavior of the higher moments of the graph-based test statistics. We show that, under mild conditions of the graph, tightness of the stochastic process can be established by obtaining upper bounds on the graph-based statistics' higher moments. Explicit analytical expressions for these moments are provided. The results are applicable to generic graphs, including dense graphs where the number of edges can be of higher order than the number of observations.

math.PR↗

arXiv2023

Hierarchical Cross-modal Transformer for RGB-D Salient Object Detection

Most of existing RGB-D salient object detection (SOD) methods follow the CNN-based paradigm, which is unable to model long-range dependencies across space and modalities due to the natural locality of CNNs. Here we propose the Hierarchical Cross-modal Transformer (HCT), a new multi-modal transformer, to tackle this problem. Unlike previous multi-modal transformers that directly connecting all patches from two modalities, we explore the cross-modal complementarity hierarchically to respect the modality gap and spatial discrepancy in unaligned regions. Specifically, we propose to use intra-modal self-attention to explore complementary global contexts, and measure spatial-aligned inter-modal attention locally to capture cross-modal correlations. In addition, we present a Feature Pyramid module for Transformer (FPT) to boost informative cross-scale integration as well as a consistency-complementarity module to disentangle the multi-modal integration path and improve the fusion adaptivity. Comprehensive experiments on a large variety of public datasets verify the efficacy of our designs and the consistent improvement over state-of-the-art models.

cs.CV↗

arXiv2023

Joule-Thomson expansion of charged dilatonic black holes

Based on the Einstein-Maxwell theory, the Joule-Thomson (J-T) expansion of charged dilatonic black holes (the solutions are neither flat nor AdS) in $(n+1)$-dimensional spacetime is studied herein. To this end, we analyze the effects of the dimension $n$ and dilaton field $α$ on J-T expansion. An explicit expression for the J-T coefficient is derived, and consequently, a negative heat capacity is found to lead to a cooling process. In contrast to its effect on the dimension, the inversion curve decreases with charge $Q$ at low pressures, whereas the opposite effect is observed at high pressures. We can observe that with an increase in the dimension $n$ or parameter $α$, both the pressure cut-off point and the minimum inversion temperature $T_{min}$ change. Moreover, we analyze the ratio $T_{min}/T_{c}$ numerically and discover that the ratio is independent of charge; however, it depends on the dilaton field and dimension: for $n=3$ and $α=0$, the ratio is 1/2. The dilaton field is found to enhance the ratio. In addition, we identify the cooling-heating regions by investigating the inversion and isenthalpic curves, and the behavior of the minimum inversion mass $M_{min}$ indicates that this cooling-heating transition may not occur under certain special conditions.

gr-qc↗

arXiv2023

Transformation and reconstruction towards two-dimensional atomic laminates

Two-dimensional (2D) nanomaterials derived from non-van der Waals solids are promising due to their fantastic physical and chemical properties, but it remains challenging to obtain 2D atomic laminates with high stability owing to the strong intrinsic covalent/metallic bonds and highly exposed active surface. Here, we report a versatile and scalable protocol to produce 2D atomic laminates, based on an unexpected topological transformation of MAX phases under hydrogen chloride gas, and/or subsequent reconstruction under some programmed gases/vapors. In contrast to the known approaches with liquid or molten medium, our method involves in a gas-phase reaction with fast thermodynamics for A layers and positive Gibbs free energies for MX slabs. Remarkably, through subsequent reconstruction in some active gases/vapors (O2, H2S, P, CH4, Al and Sn metal vapors), a big family of 2D atomic laminates with elusive configurations as well as high chemical/thermal stabilities and tunable electrical properties (from metallic to semiconductor-like behaviors) are achieved. Moreover, the resultant 2D atomic laminates can be facilely scaled up to 10 kilograms. We believe that the 2D atomic laminates would have broad applications in catalysis, energy storage, electromagnetic shielding interface and microwave absorption.

cond-mat.mtrl-sci↗

arXiv2023

Metal-bonded Atomic Layers of Transition Metal Carbides (MXenes)

Although two-dimensional transition metal carbides and nitrides (MXenes) have fantastic physical and chemical properties as well as wide applications, it remains challenging to produce stable MXenes due to their rapid structural degradation. Here, unique metal-bonded atomic layers of transition metal carbides with high stabilities are produced via a simple topological reaction between chlorine-terminated MXenes and selected metals, where the metals enable to not only remove Cl terminations, but also efficiently bond with adjacent atomic MXene slabs, driven by the symmetry of MAX phases. The films constructed from Al-bonded Ti$_3$C$_2$Cl$_x$ atomic layers show high oxidation resistance up to 400 degrees centigrade and low sheet resistance of 9.3 ohm per square. Coupled to the multi-layer structure, the Al-bonded Ti$_3$C$_2$Cl$_x$ film displays a significantly improved EMI shielding capability with a total shielding effectiveness value of 39 dB at a low thickness of 3.1 micron, outperforming pure Ti$_3$C$_2$Cl$_x$ film.

cond-mat.mtrl-sci↗