arXiv subjects

Haotian Chen

Publications and source records attributed to Haotian Chen.

66 records · Page 4Linked to original sources

arXiv2023

Does Debiasing Inevitably Degrade the Model Performance

Gender bias in language models has attracted sufficient attention because it threatens social justice. However, most of the current debiasing methods degraded the model's performance on other tasks while the degradation mechanism is still mysterious. We propose a theoretical framework explaining the three candidate mechanisms of the language model's gender bias. We use our theoretical framework to explain why the current debiasing methods cause performance degradation. We also discover a pathway through which debiasing will not degrade the model performance. We further develop a causality-detection fine-tuning approach to correct gender bias. The numerical experiment demonstrates that our method is able to lead to double dividends: partially mitigating gender bias while avoiding performance degradation.

cs.CL↗

arXiv2023

Model-Contrastive Federated Domain Adaptation

Federated domain adaptation (FDA) aims to collaboratively transfer knowledge from source clients (domains) to the related but different target client, without communicating the local data of any client. Moreover, the source clients have different data distributions, leading to extremely challenging in knowledge transfer. Despite the recent progress in FDA, we empirically find that existing methods can not leverage models of heterogeneous domains and thus they fail to achieve excellent performance. In this paper, we propose a model-based method named FDAC, aiming to address {\bf F}ederated {\bf D}omain {\bf A}daptation based on {\bf C}ontrastive learning and Vision Transformer (ViT). In particular, contrastive learning can leverage the unlabeled data to train excellent models and the ViT architecture performs better than convolutional neural networks (CNNs) in extracting adaptable features. To the best of our knowledge, FDAC is the first attempt to learn transferable representations by manipulating the latent architecture of ViT under the federated setting. Furthermore, FDAC can increase the target data diversity by compensating from each source model with insufficient knowledge of samples and features, based on domain augmentation and semantic matching. Extensive experiments on several real datasets demonstrate that FDAC outperforms all the comparative methods in most conditions. Moreover, FDCA can also improve communication efficiency which is another key factor in the federated setting.

cs.LG↗

arXiv2023

Knowledge is Power: Understanding Causality Makes Legal judgment Prediction Models More Generalizable and Robust

Legal Judgment Prediction (LJP), aiming to predict a judgment based on fact descriptions according to rule of law, serves as legal assistance to mitigate the great work burden of limited legal practitioners. Most existing methods apply various large-scale pre-trained language models (PLMs) finetuned in LJP tasks to obtain consistent improvements. However, we discover the fact that the state-of-the-art (SOTA) model makes judgment predictions according to irrelevant (or non-casual) information. The violation of rule of law not only weakens the robustness and generalization ability of models but also results in severe social problems like discrimination. In this paper, we use causal structural models (SCMs) to theoretically analyze how LJP models learn to make decisions and why they can succeed in passing the traditional testing paradigm without learning causality. According to our analysis, we provide two solutions intervening on data and model by causality, respectively. In detail, we first distinguish non-causal information by applying the open information extraction (OIE) technique. Then, we propose a method named the Causal Information Enhanced SAmpling Method (CIESAM) to eliminate the non-causal information from data. To validate our theoretical analysis, we further propose another method using our proposed Causality-Aware Self-Attention Mechanism (CASAM) to guide the model to learn the underlying causality knowledge in legal texts. The confidence of CASAM in learning causal information is higher than that of CIESAM. The extensive experimental results show that both our proposed methods achieve state-of-the-art (SOTA) performance on three commonly used legal-specific datasets. The stronger performance of CASAM further demonstrates that causality is the key to the robustness and generalization ability of models.

cs.CL↗

arXiv2023

GAT-COBO: Cost-Sensitive Graph Neural Network for Telecom Fraud Detection

Along with the rapid evolution of mobile communication technologies, such as 5G, there has been a drastically increase in telecom fraud, which significantly dissipates individual fortune and social wealth. In recent years, graph mining techniques are gradually becoming a mainstream solution for detecting telecom fraud. However, the graph imbalance problem, caused by the Pareto principle, brings severe challenges to graph data mining. This is a new and challenging problem, but little previous work has been noticed. In this paper, we propose a Graph ATtention network with COst-sensitive BOosting (GAT-COBO) for the graph imbalance problem. First, we design a GAT-based base classifier to learn the embeddings of all nodes in the graph. Then, we feed the embeddings into a well-designed cost-sensitive learner for imbalanced learning. Next, we update the weights according to the misclassification cost to make the model focus more on the minority class. Finally, we sum the node embeddings obtained by multiple cost-sensitive learners to obtain a comprehensive node representation, which is used for the downstream anomaly detection task. Extensive experiments on two real-world telecom fraud detection datasets demonstrate that our proposed method is effective for the graph imbalance problem, outperforming the state-of-the-art GNNs and GNN-based fraud detectors. In addition, our model is also helpful for solving the widespread over-smoothing problem in GNNs. The GAT-COBO code and datasets are available at https://github.com/xxhu94/GAT-COBO.

cs.LG↗

arXiv2023

Cost Sensitive GNN-based Imbalanced Learning for Mobile Social Network Fraud Detection

With the rapid development of mobile networks, the people's social contacts have been considerably facilitated. However, the rise of mobile social network fraud upon those networks, has caused a great deal of distress, in case of depleting personal and social wealth, then potentially doing significant economic harm. To detect fraudulent users, call detail record (CDR) data, which portrays the social behavior of users in mobile networks, has been widely utilized. But the imbalance problem in the aforementioned data, which could severely hinder the effectiveness of fraud detectors based on graph neural networks(GNN), has hardly been addressed in previous work. In this paper, we are going to present a novel Cost-Sensitive Graph Neural Network (CSGNN) by creatively combining cost-sensitive learning and graph neural networks. We conduct extensive experiments on two open-source realworld mobile network fraud datasets. The results show that CSGNN can effectively solve the graph imbalance problem and then achieve better detection performance than the state-of-the-art algorithms. We believe that our research can be applied to solve the graph imbalance problems in other fields. The CSGNN code and datasets are publicly available at https://github.com/xxhu94/CSGNN.

cs.SI↗

arXiv2023

Zonal Flow Excitation in Electron-Scale Tokamak Turbulence

The derivation of an intermediate-scale gyrokinetic-electron theory in nonuniform tokamak plasmas [Chen H. et al 2021 Nucl. Fusion 61 066017] has shown that a Navier-Stokes type nonlinearity couples electron-temperature-gradient (ETG) modes and zonal flow (ZF) modes with wavelengths much shorter than the ion gyroradius but much longer than the electron gyroradius. This intermediate-scale ETG-ZF coupling is typically stronger than the Hasegawa-Mima type nonlinearity characteristic of the fluid approximation and is predicted to lead to relevant zonal flow generation and ETG mode regulation. Electron-scale, continuum, gyrokinetic simulation results are presented here which include both single-mode ETG and full-spectrum ETG turbulence. The zonal flow generation due to single ETG modes is investigated and the single-mode intermediate-scale results are found to be in agreement with theory. The full-spectrum results are then presented and explained qualitatively in terms of the single-mode results. It is found that the ETG-driven zonal flows regulate intermediate-scale electron heat flux transport to levels in the predicted range.

physics.plasm-ph↗

arXiv2022

Gyrokinetic Theory of Low-frequency Electromagnetic Waves in Finite-$β$ Anisotropic Plasmas

We present a gyrokinetic theory for the electromagnetic waves and instabilities with frequencies much lower than the ion cyclotron frequency in finite-$β$ anisotropic uniform plasmas. Here, $β$ is the ratio between plasma and magnetic pressures. Kinetic effects due to both the finite Larmor radii and wave-particle resonances are fully kept in the analysis. Corresponding linear dispersion relation and wave polarizations, valid for general $β$ value and perpendicular wavelength, are then specifically derived for a bi-Maxwellian plasma. Analytic expressions for the criteria of kinetic firehose and mirror instabilities are also given. The mode frequency, stability, and wave polarization of a broad spectrum of normal modes are then investigated numerically in a systematic study over a set of parameters. Our study clearly demonstrates that, due to the finite ion Larmor radius effect, the ion-sound wave, mirror mode, and shear Alfven wave are intrinsically coupled.

physics.plasm-ph↗

arXiv2021

How Zonal Flow Affects Trapped Electron Driven Turbulence in Tokamaks

The role of self-generated zonal flows in the collisionless trapped-electron mode (CTEM) turbulence is a long-standing open issue in tokamak plasmas. Here we show, for the first time, that the zonal flow excitation in the CTEM turbulence is formally isomorphic to that in the ion temperature gradient turbulence. Trapped electrons contribute implicitly only via linear dynamics. Theoretical analyses further suggest that, for short wavelength CTEMs, the zonal flow excitation is weak and, more importantly, not an effective saturation mechanism. Corresponding controlling parameters are also identified theoretically. These findings not only offer a plausible explanation for previous seemingly contradictory simulation results, but can also facilitate controlling the CTEM instability and transport with experimentally accessible parameters.

physics.plasm-ph↗

arXiv2021

On locating the zeros and poles of a meromorphic function

On the basis of the generalized argument principle, here we develop a numerical scheme for locating zeros and poles of a meromorphic function. A subdivision-transformation-calculation scheme is proposed to ensure the algorithm stability. A novel feature of this algorithm is the ability to estimate the error level automatically. Numerical examples are also presented, with an emphasis on potential applications to plasma physics.

math.NA↗

arXiv2020

A Gyrokinetic Simulation Model for Low Frequency Electromagnetic Fluctuations in Magnetized Plasmas

We present a new model for simulating the electromagnetic fluctuations with frequencies much lower than the ion cyclotron frequency in plasmas confined in general magnetic configurations. This novel model (termed as GK-E&B) employs nonlinear gyrokinetic equations formulated in terms of electromagnetic fields along with momentum balance equations for solving fields. It, thus, not only includes kinetic effects, such as wave-particle interaction and microscopic (ion Larmor radius scale) physics; but also is computationally more efficient than the conventional formulation described in terms of potentials. As a benchmark, we perform linear as well as nonlinear simulations of the kinetic Alfven wave; demonstrating physics in agreement with the analytical theories.

physics.plasm-ph↗

arXiv2020

Excitation of Zonal Flow by Intermediate-Scale Toroidal Electron Temperature Gradient Turbulence

We show that zonal flow can be preferentially excited by intermediate-scale toroidal electron temperature gradient (ETG) turbulence in tokamak plasmas. Previous theoretical studies that yielded an opposite conclusion assumed a fluid approximation for ETG modes. Here, we carry out a gyrokinetic analysis which ultimately yields a nonlinear Schrödinger equation for the ETG dynamics with a Navier-Stokes type nonlinearity. For typical tokamak parameters, it is found that zonal flow generation plays an important role in the intermediate-scale ETG turbulence. This finding offers an explanation for recent multi-scale gyrokinetic simulations.

physics.plasm-ph↗

arXiv2020

Contextualized End-to-End Neural Entity Linking

We propose yet another entity linking model (YELM) which links words to entities instead of spans. This overcomes any difficulties associated with the selection of good candidate mention spans and makes the joint training of mention detection (MD) and entity disambiguation (ED) easily possible. Our model is based on BERT and produces contextualized word embeddings which are trained against a joint MD and ED objective. We achieve state-of-the-art results on several standard entity linking (EL) datasets.

cs.CL↗