arXiv subjects

Shiyuan Li

Publications and source records attributed to Shiyuan Li.

26 records · Page 2Linked to original sources

arXiv2024

ARC: A Generalist Graph Anomaly Detector with In-Context Learning

Graph anomaly detection (GAD), which aims to identify abnormal nodes that differ from the majority within a graph, has garnered significant attention. However, current GAD methods necessitate training specific to each dataset, resulting in high training costs, substantial data requirements, and limited generalizability when being applied to new datasets and domains. To address these limitations, this paper proposes ARC, a generalist GAD approach that enables a ``one-for-all'' GAD model to detect anomalies across various graph datasets on-the-fly. Equipped with in-context learning, ARC can directly extract dataset-specific patterns from the target dataset using few-shot normal samples at the inference stage, without the need for retraining or fine-tuning on the target dataset. ARC comprises three components that are well-crafted for capturing universal graph anomaly patterns: 1) smoothness-based feature Alignment module that unifies the features of different datasets into a common and anomaly-sensitive space; 2) ego-neighbor Residual graph encoder that learns abnormality-related node embeddings; and 3) cross-attentive in-Context anomaly scoring module that predicts node abnormality by leveraging few-shot normal samples. Extensive experiments on multiple benchmark datasets from various domains demonstrate the superior anomaly detection performance, efficiency, and generalizability of ARC.

cs.LG↗

arXiv2024

Noise-Resilient Unsupervised Graph Representation Learning via Multi-Hop Feature Quality Estimation

Unsupervised graph representation learning (UGRL) based on graph neural networks (GNNs), has received increasing attention owing to its efficacy in handling graph-structured data. However, existing UGRL methods ideally assume that the node features are noise-free, which makes them fail to distinguish between useful information and noise when applied to real data with noisy features, thus affecting the quality of learned representations. This urges us to take node noisy features into account in real-world UGRL. With empirical analysis, we reveal that feature propagation, the essential operation in GNNs, acts as a "double-edged sword" in handling noisy features - it can both denoise and diffuse noise, leading to varying feature quality across nodes, even within the same node at different hops. Building on this insight, we propose a novel UGRL method based on Multi-hop feature Quality Estimation (MQE for short). Unlike most UGRL models that directly utilize propagation-based GNNs to generate representations, our approach aims to learn representations through estimating the quality of propagated features at different hops. Specifically, we introduce a Gaussian model that utilizes a learnable "meta-representation" as a condition to estimate the expectation and variance of multi-hop propagated features via neural networks. In this way, the "meta representation" captures the semantic and structural information underlying multiple propagated features but is naturally less susceptible to interference by noise, thereby serving as high-quality node representations beneficial for downstream tasks. Extensive experiments on multiple real-world datasets demonstrate that MQE in learning reliable node representations in scenarios with diverse types of feature noise.

cs.LG↗

arXiv2022

aSTDP: A More Biologically Plausible Learning

Spike-timing dependent plasticity in biological neural networks has been proven to be important during biological learning process. On the other hand, artificial neural networks use a different way to learn, such as Back-Propagation or Contrastive Hebbian Learning. In this work we introduce approximate STDP, a new neural networks learning framework more similar to the biological learning process. It uses only STDP rules for supervised and unsupervised learning, every neuron distributed learn patterns and don' t need a global loss or other supervised information. We also use a numerical way to approximate the derivatives of each neuron in order to better use SDTP learning and use the derivatives to set a target for neurons to accelerate training and testing process. The framework can make predictions or generate patterns in one model without additional configuration. Finally, we verified our framework on MNIST dataset for classification and generation tasks.

cs.NE↗

arXiv2021

MSTDP: A More Biologically Plausible Learning

Spike-timing dependent plasticity (STDP) which observed in the brain has proven to be important in biological learning. On the other hand, artificial neural networks use a different way to learn, such as Back-Propagation or Contrastive Hebbian Learning. In this work, we propose a new framework called mstdp that learn almost the same way biological learning use, it only uses STDP rules for supervised and unsupervised learning and don' t need a global loss or other supervise information. The framework works like an auto-encoder by making each input neuron also an output neuron. It can make predictions or generate patterns in one model without additional configuration. We also brought a new iterative inference method using momentum to make the framework more efficient, which can be used in training and testing phases. Finally, we verified our framework on MNIST dataset for classification and generation task.

cs.NE↗

arXiv2017

Dynamics of coupled phantom and tachyon fields

In this paper, we apply the dynamical analysis to a coupled phantom field with scaling potential taking particular forms of the coupling (linear and combination of linear), and present phase space analysis. We investigate if there exist late time accelerated scaling attractor that has the ratio of dark energy and dark matter densities of the order one. We observe that the scrutinized couplings cannot alleviate the coincidence problem, however acquire stable late time accelerated solutions. We also discuss coupled tachyon field with inverse square potential assuming linear coupling.

gr-qc↗

arXiv2017

Rotation Invariance Neural Network

Rotation invariance and translation invariance have great values in image recognition tasks. In this paper, we bring a new architecture in convolutional neural network (CNN) named cyclic convolutional layer to achieve rotation invariance in 2-D symbol recognition. We can also get the position and orientation of the 2-D symbol by the network to achieve detection purpose for multiple non-overlap target. Last but not least, this architecture can achieve one-shot learning in some cases using those invariance.

cs.CV↗

arXiv2016

Thermodynamics of interacting tachyonic scalar field

In this paper we discuss the laws of thermodynamics for interacting tachyonic scalar field. The components of the tachyonic scalar field in the universe are taken to exist in the state of non-equilibrium initially, but due to interaction they undergo a transition towards the equilibrium state. We show that the zeroth law of thermodynamics demands interaction among the components of cosmic field. The second law of thermodynamics is governing dynamics in transfer of energy among the three components of the proposed field with local violation of conservation of energy for individual components.

gr-qc↗

arXiv2013

Combination of e+/e- ratio from AMS-02 and gamma ray line from Fermi-LAT with implication for Dark Matter

The precise AMS-02 data provide definite information for the e+/e- ratio in 100 - 350 GeV region. Assuming that the recent gamma ray line observed by Fermi-LAT experiment is product of dark matter in space and taken as input. We make a global fit for the AMS-02 e+/e- spectrum with both dark matter and pulsar contribution. For this spectrum over 130 GeV, pulsar is the dominant contribution. We also investigate the constrains on hadron production from dark matter annihilation.

hep-ph↗