arXiv subjects

Shang Gao

Publications and source records attributed to Shang Gao.

At least 73 records · Page 4Linked to original sources

arXiv2023

Part Representation Learning with Teacher-Student Decoder for Occluded Person Re-identification

Occluded person re-identification (ReID) is a very challenging task due to the occlusion disturbance and incomplete target information. Leveraging external cues such as human pose or parsing to locate and align part features has been proven to be very effective in occluded person ReID. Meanwhile, recent Transformer structures have a strong ability of long-range modeling. Considering the above facts, we propose a Teacher-Student Decoder (TSD) framework for occluded person ReID, which utilizes the Transformer decoder with the help of human parsing. More specifically, our proposed TSD consists of a Parsing-aware Teacher Decoder (PTD) and a Standard Student Decoder (SSD). PTD employs human parsing cues to restrict Transformer's attention and imparts this information to SSD through feature distillation. Thereby, SSD can learn from PTD to aggregate information of body parts automatically. Moreover, a mask generator is designed to provide discriminative regions for better ReID. In addition, existing occluded person ReID benchmarks utilize occluded samples as queries, which will amplify the role of alleviating occlusion interference and underestimate the impact of the feature absence issue. Contrastively, we propose a new benchmark with non-occluded queries, serving as a complement to the existing benchmark. Extensive experiments demonstrate that our proposed method is superior and the new benchmark is essential. The source codes are available at https://github.com/hh23333/TSD.

cs.CV↗

arXiv2023

Robust Backdoor Detection for Deep Learning via Topological Evolution Dynamics

A backdoor attack in deep learning inserts a hidden backdoor in the model to trigger malicious behavior upon specific input patterns. Existing detection approaches assume a metric space (for either the original inputs or their latent representations) in which normal samples and malicious samples are separable. We show that this assumption has a severe limitation by introducing a novel SSDT (Source-Specific and Dynamic-Triggers) backdoor, which obscures the difference between normal samples and malicious samples. To overcome this limitation, we move beyond looking for a perfect metric space that would work for different deep-learning models, and instead resort to more robust topological constructs. We propose TED (Topological Evolution Dynamics) as a model-agnostic basis for robust backdoor detection. The main idea of TED is to view a deep-learning model as a dynamical system that evolves inputs to outputs. In such a dynamical system, a benign input follows a natural evolution trajectory similar to other benign inputs. In contrast, a malicious sample displays a distinct trajectory, since it starts close to benign samples but eventually shifts towards the neighborhood of attacker-specified target samples to activate the backdoor. Extensive evaluations are conducted on vision and natural language datasets across different network architectures. The results demonstrate that TED not only achieves a high detection rate, but also significantly outperforms existing state-of-the-art detection approaches, particularly in addressing the sophisticated SSDT attack. The code to reproduce the results is made public on GitHub.

cs.CR↗

arXiv2023

Enhanced Knowledge Injection for Radiology Report Generation

Automatic generation of radiology reports holds crucial clinical value, as it can alleviate substantial workload on radiologists and remind less experienced ones of potential anomalies. Despite the remarkable performance of various image captioning methods in the natural image field, generating accurate reports for medical images still faces challenges, i.e., disparities in visual and textual data, and lack of accurate domain knowledge. To address these issues, we propose an enhanced knowledge injection framework, which utilizes two branches to extract different types of knowledge. The Weighted Concept Knowledge (WCK) branch is responsible for introducing clinical medical concepts weighted by TF-IDF scores. The Multimodal Retrieval Knowledge (MRK) branch extracts triplets from similar reports, emphasizing crucial clinical information related to entity positions and existence. By integrating this finer-grained and well-structured knowledge with the current image, we are able to leverage the multi-source knowledge gain to ultimately facilitate more accurate report generation. Extensive experiments have been conducted on two public benchmarks, demonstrating that our method achieves superior performance over other state-of-the-art methods. Ablation studies further validate the effectiveness of two extracted knowledge sources.

cs.CV↗

arXiv2023

LegalBench: A Collaboratively Built Benchmark for Measuring Legal Reasoning in Large Language Models

The advent of large language models (LLMs) and their adoption by the legal community has given rise to the question: what types of legal reasoning can LLMs perform? To enable greater study of this question, we present LegalBench: a collaboratively constructed legal reasoning benchmark consisting of 162 tasks covering six different types of legal reasoning. LegalBench was built through an interdisciplinary process, in which we collected tasks designed and hand-crafted by legal professionals. Because these subject matter experts took a leading role in construction, tasks either measure legal reasoning capabilities that are practically useful, or measure reasoning skills that lawyers find interesting. To enable cross-disciplinary conversations about LLMs in the law, we additionally show how popular legal frameworks for describing legal reasoning -- which distinguish between its many forms -- correspond to LegalBench tasks, thus giving lawyers and LLM developers a common vocabulary. This paper describes LegalBench, presents an empirical evaluation of 20 open-source and commercial LLMs, and illustrates the types of research explorations LegalBench enables.

cs.CL↗

arXiv2023

Spinon continuum in the Heisenberg quantum chain compound Sr$_2$V$_3$O$_9$

Magnetic excitations in the spin chain candidate Sr$_2$V$_3$O$_9$ have been investigated by inelastic neutron scattering on a single crystal sample. A spinon continuum with a bandwidth of $\sim22$ meV is observed along the chain formed by alternating magnetic V$^{4+}$ and nonmagnetic V$^{5+}$ ions. Incipient magnetic Bragg peaks due to weak ferromagnetic interchain couplings emerge when approaching the magnetic transition at $T_N\sim 5.3$ K while the excitations remain gapless within the instrumental resolution. Comparisons to the Bethe ansatz, density matrix renormalization group (DMRG) calculations, and effective field theories confirm Sr$_2$V$_3$O$_9$ as a host of weakly coupled $S = 1/2$ chains dominated by antiferromagnetic intrachain interactions of $\sim7.1$(1) meV.

cond-mat.str-el↗

arXiv2023

Can Pretrained Language Models Derive Correct Semantics from Corrupt Subwords under Noise?

For Pretrained Language Models (PLMs), their susceptibility to noise has recently been linked to subword segmentation. However, it is unclear which aspects of segmentation affect their understanding. This study assesses the robustness of PLMs against various disrupted segmentation caused by noise. An evaluation framework for subword segmentation, named Contrastive Lexical Semantic (CoLeS) probe, is proposed. It provides a systematic categorization of segmentation corruption under noise and evaluation protocols by generating contrastive datasets with canonical-noisy word pairs. Experimental results indicate that PLMs are unable to accurately compute word meanings if the noise introduces completely different subwords, small subword fragments, or a large number of additional subwords, particularly when they are inserted within other subwords.

cs.CL↗

arXiv2023

A Survey on Out-of-Distribution Evaluation of Neural NLP Models

Adversarial robustness, domain generalization and dataset biases are three active lines of research contributing to out-of-distribution (OOD) evaluation on neural NLP models. However, a comprehensive, integrated discussion of the three research lines is still lacking in the literature. In this survey, we 1) compare the three lines of research under a unifying definition; 2) summarize the data-generating processes and evaluation protocols for each line of research; and 3) emphasize the challenges and opportunities for future work.

cs.CL↗

arXiv2023

Track Anything: Segment Anything Meets Videos

Recently, the Segment Anything Model (SAM) gains lots of attention rapidly due to its impressive segmentation performance on images. Regarding its strong ability on image segmentation and high interactivity with different prompts, we found that it performs poorly on consistent segmentation in videos. Therefore, in this report, we propose Track Anything Model (TAM), which achieves high-performance interactive tracking and segmentation in videos. To be detailed, given a video sequence, only with very little human participation, i.e., several clicks, people can track anything they are interested in, and get satisfactory results in one-pass inference. Without additional training, such an interactive design performs impressively on video object tracking and segmentation. All resources are available on {https://github.com/gaomingqi/Track-Anything}. We hope this work can facilitate related research.

cs.CV↗

arXiv2023

Anomaly Detection of UAV State Data Based on Single-class Triangular Global Alignment Kernel Extreme Learning Machine

Unmanned Aerial Vehicles (UAVs) are widely used and meet many demands in military and civilian fields. With the continuous enrichment and extensive expansion of application scenarios, the safety of UAVs is constantly being challenged. To address this challenge, we propose algorithms to detect anomalous data collected from drones to improve drone safety. We deployed a one-class kernel extreme learning machine (OCKELM) to detect anomalies in drone data. By default, OCKELM uses the radial basis (RBF) kernel function as the kernel function of the model. To improve the performance of OCKELM, we choose a Triangular Global Alignment Kernel (TGAK) instead of an RBF Kernel and introduce the Fast Independent Component Analysis (FastICA) algorithm to reconstruct UAV data. Based on the above improvements, we create a novel anomaly detection strategy FastICA-TGAK-OCELM. The method is finally validated on the UCI dataset and detected on the Aeronautical Laboratory Failures and Anomalies (ALFA) dataset. The experimental results show that compared with other methods, the accuracy of this method is improved by more than 30%, and point anomalies are effectively detected.

cs.AI↗

arXiv2023

Cubic Double Perovskites Host Noncoplanar Spin Textures

Magnetic materials with noncoplanar magnetic structures can show unusual physical properties driven by nontrivial topology. Topologically-active states are often multi-q structures, which are challenging to stabilize in models and to identify in materials. Here, we use inelastic neutron-scattering experiments to show that the insulating double perovskites Ba2YRuO6 and Ba2LuRuO6 host a noncoplanar 3-q structure on the face-centered cubic lattice. Quantitative analysis of our neutron-scattering data reveals that these 3-q states are stabilized by biquadratic interactions. Our study identifies double perovskites as a highly promising class of materials to realize topological magnetism, elucidates the stabilization mechanism of the 3-q state in these materials, and establishes neutron spectroscopy on powder samples as a valuable technique to distinguish multi-q from single-q states, facilitating the discovery of topologically-nontrivial magnetic materials.

cond-mat.str-el↗

arXiv2022

DGD-cGAN: A Dual Generator for Image Dewatering and Restoration

Underwater images are usually covered with a blue-greenish colour cast, making them distorted, blurry or low in contrast. This phenomenon occurs due to the light attenuation given by the scattering and absorption in the water column. In this paper, we present an image enhancement approach for dewatering which employs a conditional generative adversarial network (cGAN) with two generators. Our Dual Generator Dewatering cGAN (DGD-cGAN) removes the haze and colour cast induced by the water column and restores the true colours of underwater scenes whereby the effects of various attenuation and scattering phenomena that occur in underwater images are tackled by the two generators. The first generator takes at input the underwater image and predicts the dewatered scene, while the second generator learns the underwater image formation process by implementing a custom loss function based upon the transmission and the veiling light components of the image formation model. Our experiments show that DGD-cGAN consistently delivers a margin of improvement as compared with the state-of-the-art methods on several widely available datasets.

eess.IV↗

arXiv2022

Learning Dual-Fused Modality-Aware Representations for RGBD Tracking

With the development of depth sensors in recent years, RGBD object tracking has received significant attention. Compared with the traditional RGB object tracking, the addition of the depth modality can effectively solve the target and background interference. However, some existing RGBD trackers use the two modalities separately and thus some particularly useful shared information between them is ignored. On the other hand, some methods attempt to fuse the two modalities by treating them equally, resulting in the missing of modality-specific features. To tackle these limitations, we propose a novel Dual-fused Modality-aware Tracker (termed DMTracker) which aims to learn informative and discriminative representations of the target objects for robust RGBD tracking. The first fusion module focuses on extracting the shared information between modalities based on cross-modal attention. The second aims at integrating the RGB-specific and depth-specific information to enhance the fused features. By fusing both the modality-shared and modality-specific information in a modality-aware scheme, our DMTracker can learn discriminative representations in complex tracking scenes. Experiments show that our proposed tracker achieves very promising results on challenging RGBD benchmarks.

cs.CV↗

arXiv2022

Line-Graph Approach to Spiral Spin Liquids

Competition among exchange interactions is able to induce novel spin correlations on a bipartite lattice without geometrical frustration. A prototype example is the spiral spin liquid, which is a correlated paramagnetic state characterized by sub-dimensional degenerate propagation vectors. Here, using spectral graph theory, we show that spiral spin liquids on a bipartite lattice can be approximated by a further-neighbor model on the corresponding line-graph lattice that is non-bipartite, thus broadening the space of candidate materials that may support the spiral spin liquid phases. As illustrations, we examine neutron scattering experiments performed on two spinel compounds, ZnCr$_2$Se$_4$ and CuInCr$_4$Se$_8$, to demonstrate the feasibility of this new approach and expose its possible limitations in experimental realizations.

cond-mat.str-el↗

arXiv2022

Graph Classification via Discriminative Edge Feature Learning

Spectral graph convolutional neural networks (GCNNs) have been producing encouraging results in graph classification tasks. However, most spectral GCNNs utilize fixed graphs when aggregating node features, while omitting edge feature learning and failing to get an optimal graph structure. Moreover, many existing graph datasets do not provide initialized edge features, further restraining the ability of learning edge features via spectral GCNNs. In this paper, we try to address this issue by designing an edge feature scheme and an add-on layer between every two stacked graph convolution layers in GCNN. Both are lightweight while effective in filling the gap between edge feature learning and performance enhancement of graph classification. The edge feature scheme makes edge features adapt to node representations at different graph convolution layers. The add-on layers help adjust the edge features to an optimal graph structure. To test the effectiveness of our method, we take Euclidean positions as initial node features and extract graphs with semantic information from point cloud objects. The node features of our extracted graphs are more scalable for edge feature learning than most existing graph datasets (in one-hot encoded label format). Three new graph datasets are constructed based on ModelNet40, ModelNet10 and ShapeNet Part datasets. Experimental results show that our method outperforms state-of-the-art graph classification methods on the new datasets by reaching 96.56% overall accuracy on Graph-ModelNet40, 98.79% on Graph-ModelNet10 and 97.91% on Graph-ShapeNet Part. The constructed graph datasets will be released to the community.

cs.CV↗

arXiv2022

CREAM: Weakly Supervised Object Localization via Class RE-Activation Mapping

Weakly Supervised Object Localization (WSOL) aims to localize objects with image-level supervision. Existing works mainly rely on Class Activation Mapping (CAM) derived from a classification model. However, CAM-based methods usually focus on the most discriminative parts of an object (i.e., incomplete localization problem). In this paper, we empirically prove that this problem is associated with the mixup of the activation values between less discriminative foreground regions and the background. To address it, we propose Class RE-Activation Mapping (CREAM), a novel clustering-based approach to boost the activation values of the integral object regions. To this end, we introduce class-specific foreground and background context embeddings as cluster centroids. A CAM-guided momentum preservation strategy is developed to learn the context embeddings during training. At the inference stage, the re-activation mapping is formulated as a parameter estimation problem under Gaussian Mixture Model, which can be solved by deriving an unsupervised Expectation-Maximization based soft-clustering algorithm. By simply integrating CREAM into various WSOL approaches, our method significantly improves their performance. CREAM achieves the state-of-the-art performance on CUB, ILSVRC and OpenImages benchmark datasets. Code will be available at https://github.com/Jazzcharles/CREAM.

cs.CV↗

arXiv2022

Fault Diagnosis of Discrete-Event Systems under Non-Deterministic Observations with Output Fairness

In this paper, we revisit the fault diagnosis problem of discrete-event systems (DES) under non-deterministic observations. Non-deterministic observation is a general observation model that includes the case of intermittent loss of observations. In this setting, upon the occurrence of an event, the sensor reading may be non-deterministic such that a set of output symbols are all possible. Existing works on fault diagnosis under non-deterministic observations require to consider all possible observation realizations. However, this approach includes the case where some possible outputs are permanently disabled. In this work, we introduce the concept of output fairness by requiring that, for any output symbols, if it has infinite chances to be generated, then it will indeed be generated infinite number of times. We use an assume-guarantee type of linear temporal logic formulas to formally describe this assumption. A new notion called output-fair diagnosability (OF-diagnosability) is proposed. An effective approach is provided for the verification of OF-diagnosability. We show that the proposed notion of OF-diagnosability is weaker than the standard definition of diagnosability under non-deterministic observations, and it better captures the physical scenario of observation non-determinism or intermittent loss of observations.

eess.SY↗

arXiv2022

Electronic structure, magnetic properties and pairing tendencies of the copper-based honeycomb lattice Na$_2$Cu$_2$TeO$_6$

Spin-$1/2$ chains with alternating antiferromagnetic and ferromagnetic couplings have attracted considerable interest due to the topological character of their spin excitations. Here, using density functional theory and density matrix renormalization group methods, we have systematically studied the dimerized chain system Na$_2$Cu$_2$TeO$_6$. Near the Fermi level, the dominant states are mainly contributed by the Cu $3d_{x^2-y^2}$ orbitals highly hybridized with the O $2p$ orbitals in the nonmagnetic phase, leading to an "effective" single-orbital low-energy model. Furthermore, the bandwidth of the Cu $3d_{x^2-y^2}$ states is small ($\sim 0.8$ eV), suggesting that electronic correlations will strongly affect this system. By introducing such electronic correlations, we found this system is a Mott insulator. Moreover, by calculating the magnetic exchange interactions ($J_1$, $J_2$ and $J_3$), we explained the size and sign of the exchange interactions in Na$_2$Cu$_2$TeO$_6$, in agreement with neutron experiments. In addition, we constructed a single-orbital Hubbard model for this dimerized chain system, where the quantum fluctuations are taken into account. Both AFM and FM coupling ($\uparrow$-$\downarrow$-$\downarrow$-$\uparrow$) along the chain were found in our DMRG and Lanczos calculations, in agreement with DFT and neutron results. We also calculated the hole pairing binding energy $ΔE$ which becomes negative at Hubbard $U \sim 11$ eV, indicating incipient pairing tendencies. Finally, we also looked at various cases of hole doping that always exhibit tight pairs. Thus, we believe our results for Na$_2$Cu$_2$TeO$_6$ could provide guidance to experimentalists and theorists working on this dimerized chain system, such as short-range magnetic coupling, doping effects, and possible pairing tendencies.

cond-mat.str-el↗

arXiv2022

Spiral spin-liquid on a honeycomb lattice

Spiral spin-liquids are correlated paramagnetic states with degenerate propagation vectors forming a continuous ring or surface in reciprocal space. On the honeycomb lattice, spiral spin-liquids present a novel route to realize emergent fracton excitations, quantum spin liquids, and topological spin textures, yet experimental realizations remain elusive. Here, using neutron scattering, we show that a spiral spin-liquid is realized in the van der Waals honeycomb magnet FeCl$_3$. A continuous ring of scattering is directly observed, which indicates the emergence of an approximate U(1) symmetry in momentum space. Our work demonstrates that spiral spin-liquids can be achieved in two-dimensional systems and provides a promising platform to study the fracton physics in spiral spin-liquids.

cond-mat.str-el↗