arXiv subjects
Publications and source records attributed to Chen Yang.
Magnetic Er3+ is doped into the well-studied frustrated normal spinel ZnCr2O4. Various spectroscopies are employed to prove that Er3+ successfully enters the spinel to form ZnCr2-xErxO4 for x less than 0.005. The low levels of Er3+ doping possible nonetheless have a significant effect on the frustrated magnetism and the ordering that is seen near 12 K in the undoped material.
In this paper, we introduce CrimeGraphNet, a novel approach for link prediction in criminal networks utilizingGraph Convolutional Networks (GCNs). Criminal networks are intricate and dynamic, with covert links that are challenging to uncover. Accurate prediction of these links can aid in proactive crime prevention and investigation. Existing methods often fail to capture the complex interconnections in such networks. They also struggle in scenarios where only limited labeled data is available for training. To address these challenges, we propose CrimeGraphNet, which leverages the power of GCNs for link prediction in these networks. The GCNmodel effectively captures topological features and node characteristics, making it well-suited for this task. We evaluate CrimeGraphNet on several real-world criminal network datasets. Our results demonstrate that CrimeGraphNet outperforms existing methods in terms of prediction accuracy, robustness, and computational efAciency. Furthermore, our approach enables the extraction of meaningful insights from the predicted links, thereby contributing to a better understanding of the underlying criminal activities. Overall, CrimeGraphNet represents a signiAcant step forward in the use of deep learning for criminal network analysis.
In this paper, we present CrimeGAT, a novel application of Graph Attention Networks (GATs) for predictive policing in criminal networks. Criminal networks pose unique challenges for predictive analytics due to their complex structure, multi-relational links, and dynamic behavior. Traditional methods often fail to capture these complexities, leading to suboptimal predictions. To address these challenges, we propose the use of GATs, which can effectively leverage both node features and graph structure to make predictions. Our proposed CrimeGAT model integrates attention mechanisms to weigh the importance of a node's neighbors, thereby capturing the local and global structures of criminal networks. We formulate the problem as learning a function that maps node features and graph structure to a prediction of future criminal activity. The experimental results on real-world datasets demonstrate that CrimeGAT out-performs conventional methods in predicting criminal activities, thereby providing a powerful tool for law enforcement agencies to proactively deploy resources. Furthermore, the interpretable nature of the attentionmechanism inGATs offers insights into the key players and relationships in criminal networks. This research opens new avenues for applying deep learning techniques in the Aeld of predictive policing and criminal network analysis.
In this paper, we introduce CrimeGNN, a novel application of Graph Neural Networks (GNNs) specifically designed to uncover hidden communities within criminal networks. As criminal activities increasingly rely on complex network structures, traditional methods of network analysis often fall short in detecting the intricate and dynamic communities within these networks. Leveraging the power of GNNs, CrimeGNN provides an advanced and specialized solution to this problem. The model ingests a graph structure of a criminal network, where vertices represent individuals and edges represent relationships between them. CrimeGNN aims to identify a partition of the vertex set, such that each subset represents a distinct community within the network, maximizing the modularity function. Experimental results on several benchmark datasets demonstrate the effectiveness of CrimeGNN, outperforming existing methods in terms of both accuracy and computational efficiency. The proposed framework offers significant potential for aiding law enforcement agencies in proactive policing and crime prevention measures by providing a more in-depth understanding of the structure and operation of criminal networks.
We perform the linear analysis of causality and stability for a minimal extended spin hydrodynamics up to second order of the gradient expansion. The first order spin hydrodynamics, with a rank-3 spin tensor being antisymmetric for only the last two indices, are proved to be acausal and unstable. We then consider the minimal causal spin hydrodynamics up to second order of the gradient expansion. We derive the necessary causality and stability conditions for this minimal causal spin hydrodynamics. Interestingly, the satisfaction of the stability conditions relies on the equations of state for the spin density and chemical potentials. Moreover, different with the conventional relativistic dissipative hydrodynamics, the stability of the theory seems to be broken at the finite wave-vector when the stability conditions are fulfilled at small and large wave-vector limits. It implies that the behavior in small and large wave-vector limits may be insufficient to determine the stability conditions for spin hydrodynamics in linear mode analysis.
This paper presents TransCrimeNet, a novel transformer-based model for predicting future crimes in criminal networks from textual data. Criminal network analysis has become vital for law enforcement agencies to prevent crimes. However, existing graph-based methods fail to effectively incorporate crucial textual data like social media posts and interrogation transcripts that provide valuable insights into planned criminal activities. To address this limitation, we develop TransCrimeNet which leverages the representation learning capabilities of transformer models like BERT to extract features from unstructured text data. These text-derived features are fused with graph embeddings of the criminal network for accurate prediction of future crimes. Extensive experiments on real-world criminal network datasets demonstrate that TransCrimeNet outperforms previous state-of-the-art models by 12.7\% in F1 score for crime prediction. The results showcase the benefits of combining textual and graph-based features for actionable insights to disrupt criminal enterprises.
Context: Modern software systems (e.g., Apache Spark) are usually written in multiple programming languages (PLs). There is little understanding on the phenomenon of multi-programming-language commits (MPLCs), which involve modified source files written in multiple PLs. Objective: This work aims to explore MPLCs and their impacts on development difficulty and software quality. Methods: We performed an empirical study on eighteen non-trivial Apache projects with 197,566 commits. Results: (1) the most commonly used PL combination consists of all the four PLs, i.e., C/C++, Java, JavaScript, and Python; (2) 9% of the commits from all the projects are MPLCs, and the proportion of MPLCs in 83% of the projects goes to a relatively stable level; (3) more than 90% of the MPLCs from all the projects involve source files in two PLs; (4) the change complexity of MPLCs is significantly higher than that of non-MPLCs; (5) issues fixed in MPLCs take significantly longer to be resolved than issues fixed in non-MPLCs in 89% of the projects; (6) MPLCs do not show significant effects on issue reopen; (7) source files undergoing MPLCs tend to be more bug-prone; and (8) MPLCs introduce more bugs than non-MPLCs. Conclusions: MPLCs are related to increased development difficulty and decreased software quality.
In the past decade, photonic integrated circuits (PICs) based on thin-film lithium niobate (TFLN) have advanced in various fields, including optical communication, nonlinear photonics, and quantum optics. A critical component is an efficient edge coupler connecting PICs to light sources or detectors. Here, we propose an innovative edge coupler design with a wedge-shaped TFLN waveguide and a silicon oxynitride (SiON) cladding. Experimental results show that the coupling loss between the TFLN PIC and a 3-μm mode field diameter (MFD) lensed fiber is low at 1.52 dB/facet, with the potential for improvement to 0.43 dB/facet theoretically. The coupling loss between the edge coupler and a UHNA7 fiber with an MFD of 3.2 μm is reduced to 0.92 dB/facet. This design maintains robust fabrication and alignment tolerance. Importantly, the minimum linewidth of the TFLN waveguide of the coupler (600 nm) can be easily achieved using foundry-available i-line stepper lithography. This work benefits the development of TFLN integrated platforms, such as on-chip electro-optic modulators, frequency comb generation, and quantum sensors.
The hexagonal perovskite BaNiO3, prepared via non-ceramic approaches, is known to act as a good catalyst for the oxygen-evolution reaction (OER) in alkaline media. Here we report our observation that BaNiO3 synthesized via KOH flux growth and high O2 pressure ceramic synthesis have different magnetic properties. We show that this is because the KOH flux-grown crystals made in open-air are actually a hydroxide-containing form of BaNiO3 that can be dried upon annealing in O2 flow. This work not only unveils a previously unknown aspect of the BaNiO3 OER catalyst and offers some insights into the underlying mechanism, but also suggests that hydroxide ions may be present in other hexagonal perovskite oxides prepared in wet conditions.
We report the synthesis and magnetic characterization of stuffed rare earth gallium garnets, RE3+xGa5-xO12 (RE=Lu, Yb, Er, Dy, Gd), for x up to 0.5. The excess rare earth ions partly fill the octahedral sites normally fully occupied by Ga3+, forming disordered pairs of corner-shared face-sharing magnetic tetrahedra. The Curie-Weiss constants and observed effective moments per rare earth are smaller than are seen for the unstuffed gallium garnets. No significant change in the field-dependent magnetization is observed but missing entropy is seen when integrating the low-temperature heat capacity to 0.5 K.
This paper presents a novel layered framework that integrates visual foundation models to improve robot manipulation tasks and motion planning. The framework consists of five layers: Perception, Cognition, Planning, Execution, and Learning. Using visual foundation models, we enhance the robot's perception of its environment, enabling more efficient task understanding and accurate motion planning. This approach allows for real-time adjustments and continual learning, leading to significant improvements in task execution. Experimental results demonstrate the effectiveness of the proposed framework in various robot manipulation tasks and motion planning scenarios, highlighting its potential for practical deployment in dynamic environments.
Code generation systems have been extensively developed in recent years to generate source code based on natural language instructions. However, despite their advancements, these systems still face robustness issues where even slightly different instructions can result in significantly different code semantics. Robustness is critical for code generation systems, as it can have significant impacts on software development, software quality, and trust in the generated code. Although existing testing techniques for general text-to-text software can detect some robustness issues, they are limited in effectiveness due to ignoring the characteristics of code generation systems. In this work, we propose a novel technique COCO to test the robustness of code generation systems. It exploits the usage scenario of code generation systems to make the original programming instruction more concrete by incorporating features known to be contained in the original code. A robust system should maintain code semantics for the concretized instruction, and COCO detects robustness inconsistencies when it does not. We evaluated COCO on eight advanced code generation systems, including commercial tools such as Copilot and ChatGPT, using two widely-used datasets. Our results demonstrate the effectiveness of COCO in testing the robustness of code generation systems, outperforming two techniques adopted from general text-to-text software testing by 466.66% and 104.02%, respectively. Furthermore, concretized instructions generated by COCO can help reduce robustness inconsistencies by 18.35% to 53.91% through fine-tuning.
A series of garnets of formula Er3+xGa5-xO12 is described, for which we report the crystal structures for both polycrystalline and single-crystal samples. The x limit in the garnet phase is between 0.5 and 0.6 under our conditions, with the Er fully occupying the normal garnet site plus half-occupying the octahedral site at x = 0.5 in place of the Ga normally present. Long-range antiferromagnetic order with spin ice-like frustration is suggested by the transition temperature (TN=0.8K) being much lower than the Curie-Weiss theta. The magnetic ordering temperature does not depend on the Er excess, but there is increasing residual entropy as the Er excess is increased, highlighting the potential for unusual magnetic behavior in this system.
Reconstructing deformable tissues from endoscopic stereo videos in robotic surgery is crucial for various clinical applications. However, existing methods relying only on implicit representations are computationally expensive and require dozens of hours, which limits further practical applications. To address this challenge, we introduce LerPlane, a novel method for fast and accurate reconstruction of surgical scenes under a single-viewpoint setting. LerPlane treats surgical procedures as 4D volumes and factorizes them into explicit 2D planes of static and dynamic fields, leading to a compact memory footprint and significantly accelerated optimization. The efficient factorization is accomplished by fusing features obtained through linear interpolation of each plane and enables using lightweight neural networks to model surgical scenes. Besides, LerPlane shares static fields, significantly reducing the workload of dynamic tissue modeling. We also propose a novel sample scheme to boost optimization and improve performance in regions with tool occlusion and large motions. Experiments on DaVinci robotic surgery videos demonstrate that LerPlane accelerates optimization by over 100$\times$ while maintaining high quality across various non-rigid deformations, showing significant promise for future intraoperative surgery applications.
This paper aims to detect the potential injury risk of the anterior cruciate ligament (ACL) by proposing an ACL potential injury risk assessment algorithm based on key points of the human body detected using computer vision technology. To obtain the key points data of the human body in each frame, OpenPose, an open source computer vision algorithm, was employed. The obtained data underwent preprocessing and were then fed into an ACL potential injury feature extraction model based on the Landing Error Evaluation System (LESS). This model extracted several important parameters, including the knee flexion angle, the trunk flexion on the sagittal plane, trunk flexion angle on the frontal plane, the ankle knee horizontal distance, and the ankle shoulder horizontal distance. Each of these features was assigned a threshold interval, and a segmented evaluation function was utilized to score them accordingly. To calculate the final score of the participant, the score values were input into a weighted scoring model designed based on the Analytic Hierarchy Process (AHP). The AHP based model takes into account the relative importance of each feature in the overall assessment. The results demonstrate that the proposed algorithm effectively detects the potential risk of ACL injury. The proposed algorithm demonstrates its effectiveness in detecting ACL injury risk, offering valuable insights for injury prevention and intervention strategies in sports and related fields. Code is available at: https://github.com/ZiyuGong-proj/Assessment-of-ACL-Injury-Risk-Based-on-Openpose
We present NeRFVS, a novel neural radiance fields (NeRF) based method to enable free navigation in a room. NeRF achieves impressive performance in rendering images for novel views similar to the input views while suffering for novel views that are significantly different from the training views. To address this issue, we utilize the holistic priors, including pseudo depth maps and view coverage information, from neural reconstruction to guide the learning of implicit neural representations of 3D indoor scenes. Concretely, an off-the-shelf neural reconstruction method is leveraged to generate a geometry scaffold. Then, two loss functions based on the holistic priors are proposed to improve the learning of NeRF: 1) A robust depth loss that can tolerate the error of the pseudo depth map to guide the geometry learning of NeRF; 2) A variance loss to regularize the variance of implicit neural representations to reduce the geometry and color ambiguity in the learning procedure. These two loss functions are modulated during NeRF optimization according to the view coverage information to reduce the negative influence brought by the view coverage imbalance. Extensive results demonstrate that our NeRFVS outperforms state-of-the-art view synthesis methods quantitatively and qualitatively on indoor scenes, achieving high-fidelity free navigation results.
The layout design of pipelines is a critical task in the construction industry. Currently, pipeline layout is designed manually by engineers, which is time-consuming and laborious. Automating and streamlining this process can reduce the burden on engineers and save time. In this paper, we propose a method for generating three-dimensional layout of pipelines based on deep reinforcement learning (DRL). Firstly, we abstract the geometric features of space to establish a training environment and define reward functions based on three constraints: pipeline length, elbow, and installation distance. Next, we collect data through interactions between the agent and the environment and train the DRL model. Finally, we use the well-trained DRL model to automatically design a single pipeline. Our results demonstrate that DRL models can complete the pipeline layout task in space in a much shorter time than traditional algorithms while ensuring high-quality layout outcomes.
In software development, due to the lack of knowledge or information, time pressure, complex context, and many other factors, various uncertainties emerge during the development process, leading to assumptions scattered in projects. Being unaware of certain assumptions can result in critical problems (e.g., system vulnerability and failures). The prerequisite of analyzing and understanding assumptions in software development is to identify and extract those assumptions with acceptable effort. In this paper, we proposed a tool (i.e., Assumption Miner) to automatically identify and extract assumptions on GitHub projects. To evaluate the applicability of Assumption Miner, we first presented an example of using the tool to mine assumptions from one large and popular deep learning framework project: the TensorFlow project on GitHub. We then conducted an evaluation of the tool. The results show that Assumption Miner can effectively identify and extract assumptions from the repositories on GitHub.