A5100456786- Topic Modeling
- Advanced Graph Neural Networks
- Time Series Analysis and Forecasting
- Natural Language Processing Techniques
- Machine Learning in Healthcare
- Explainable Artificial Intelligence (XAI)
- Domain Adaptation and Few-Shot Learning
- Advanced Text Analysis Techniques
- Anomaly Detection Techniques and Applications
- Adversarial Robustness in Machine Learning
- Recommender Systems and Techniques
- Music and Audio Processing
- Text and Document Classification Technologies
- Neural Networks and Applications
- Sentiment Analysis and Opinion Mining
- Multimodal Machine Learning Applications
- Privacy-Preserving Technologies in Data
- Complex Network Analysis Techniques
- Software System Performance and Reliability
- Speech and Audio Processing
- Face and Expression Recognition
- Human Mobility and Location-Based Analysis
- Data Stream Mining Techniques
- Artificial Intelligence in Healthcare and Education
- Traffic Prediction and Management Techniques
NEC (United States)
2020-2025
Princeton University
2007-2024
Tsinghua University
2022
NEC (Japan)
2020-2021
Georgia Institute of Technology
2019
Rutgers, The State University of New Jersey
2004
The Nonlinear autoregressive exogenous (NARX) model, which predicts the current value of a time series based upon its previous values as well and past multiple driving (exogenous) series, has been studied for decades. Despite fact that various NARX models have developed, few them can capture long-term temporal dependencies appropriately select relevant to make predictions. In this paper, we propose dual-stage attention-based recurrent neural network (DA-RNN) address these two issues. first...
Despite recent progress in Graph Neural Networks (GNNs), explaining predictions made by GNNs remains a challenging open problem. The leading method independently addresses the local explanations (i.e., important subgraph structure and node features) to interpret why GNN model makes prediction for single instance, e.g. or graph. As result, explanation generated is painstakingly customized each instance. unique interpreting instance not sufficient provide global understanding of learned model,...
Text summarization has been a crucial problem in natural language processing (NLP) for several decades. It aims to condense lengthy documents into shorter versions while retaining the most critical information. Various methods have proposed text summarization, including extractive and abstractive summarization. The emergence of large models (LLMs) like GPT3 ChatGPT recently created significant interest using these tasks. Recent studies \cite{goyal2022news, zhang2023benchmarking} shown that...
To subvert recent advances in perimeter and host security, the attacker community has developed employed various attack vectors to make a malware much stealthier than before penetrate target system prolong its presence.Such advanced or "stealthy malware" makes use of techniques impersonate abuse benign applications legitimate tools minimize footprints system.It is thus difficult for traditional detection tools, such as scanners, detect it, normally does not expose malicious payload file...
Click-through rate (CTR) prediction is a critical task in online advertising and marketing. For this problem, existing approaches, with shallow or deep architectures, have three major drawbacks. First, they typically lack persuasive rationales to explain the outcomes of models. Unexplainable predictions recommendations may be difficult validate thus unreliable untrustworthy. In many applications, inappropriate suggestions even bring severe consequences. Second, approaches poor efficiency...
Various graph contrastive learning models have been proposed to improve the performance of tasks on datasets in recent years. While effective and prevalent, these are usually carefully customized. In particular, although all researches create two views, they differ greatly view augmentations, architectures, objectives. It remains an open question how build your model from scratch for particular datasets. this work, we aim fill gap by studying information is transformed transferred during...
Large language models (LLMs) have significantly advanced the field of natural processing (NLP), providing a highly useful, task-agnostic foundation for wide range applications. However, directly applying LLMs to solve sophisticated problems in specific domains meets many hurdles, caused by heterogeneity domain data, sophistication knowledge, uniqueness objectives, and diversity constraints (e.g., various social norms, cultural conformity, religious beliefs, ethical standards applications)....
Various contrastive learning approaches have been proposed in recent years and achieve significant empirical success. While effective prevalent, has less explored for time series data. A key component of is to select appropriate augmentations imposing some priors construct feasible positive samples, such that an encoder can be trained learn robust discriminative representations. Unlike image language domains where "desired'' augmented samples generated with the rule thumb guided by...
Time series data is essential in various applications, including climate modeling, healthcare monitoring, and financial analytics. Understanding the contextual information associated with real-world time often for accurate reliable event predictions. In this paper, we introduce TimeCAP, a time-series processing framework that creatively employs Large Language Models (LLMs) as contextualizers of data, extending their typical usage predictors. TimeCAP incorporates two independent LLM agents:...
Graph Neural Networks (GNNs) resurge as a trending research subject owing to their impressive ability capture representations from graph-structured data. However, the black-box nature of GNNs presents significant challenge in terms comprehending and trusting these models, thereby limiting practical applications mission-critical scenarios. Although there has been substantial progress field explaining recent years, majority studies are centered on static graphs, leaving explanation dynamic...
Multivariate time series data are becoming increasingly common in numerous real world applications, e.g., power plant monitoring, health care, wearable devices, automobile, etc. As a result, multivariate retrieval, i.e., given the current segment, how to obtain its relevant segments historical (or database), attracts significant amount of interest many fields. Building such system, however, is challenging since it requires compact representation raw which can explicitly encode temporal...
Multi-modal fusion overcomes the inherent limitations of single-sensor perception in 3D object detection autonomous driving. The 4D Radar and LiDAR can boost range more robust. Nevertheless, different data characteristics noise distributions between two sensors hinder performance improvement when directly integrating them. Therefore, we are first to propose a novel method termed <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML"...
The problem of learning and forecasting underlying trends in time series data arises a variety applications, such as traffic management, energy optimization, etc. In literature, trend is characterized by the slope duration, its prediction then to forecast two values subsequent given historical series. For this problem, existing approaches mainly deal with case univariate However, many real-world there are multiple variables at play, handling all them same crucial for an accurate prediction....
Forecasting on sparse multivariate time series (MTS) aims to model the predictors of future values given their incomplete past, which is important for many emerging applications. However, most existing methods process MTS’s individually, and do not leverage dynamic distributions underlying MTS’s, leading sub-optimal results when sparsity high. To address this challenge, we propose a novel generative model, tracks transition latent clusters, instead isolated feature representations, achieve...
The goal of root cause analysis is to identify the underlying causes system problems by discovering and analyzing causal structure from monitoring data. It indispensable for maintaining stability robustness large-scale complex systems. Existing methods mainly focus on construction a single effective isolated network, whereas many real-world systems are exhibit interdependent structures (i.e., multiple networks interconnected cross-network links). In networks, malfunctioning effects...
Deep learning systems on the cloud are increasingly targeted by attacks that attempt to steal sensitive data. Intel SGX has been proven effective protect confidentiality and integrity of such data during computation. However, state-of-the-art still suffer from substantial performance overhead induced limited physical memory SGX. This limitation significantly undermines usability deep due their memory-intensive characteristics.
The task of root cause analysis (RCA) is to identify the causes system faults/failures by analyzing monitoring data. Efficient RCA can greatly accelerate failure recovery and mitigate damages or financial losses. However, previous research has mostly focused on developing offline algorithms, which often require manually initiating process, a significant amount time data train robust model, then being retrained from scratch for new fault.
Time series analysis has witnessed the inspiring development from traditional autoregressive models, deep learning to recent Transformers and Large Language Models (LLMs). Efforts in leveraging vision models for time have also been made along way but are less visible community due predominant research on sequence modeling this domain. However, discrepancy between continuous discrete token space of LLMs, challenges explicitly correlations variates multivariate shifted some attentions equally...
Time series data is essential in various applications, including climate modeling, healthcare monitoring, and financial analytics. Understanding the contextual information associated with real-world time often for accurate reliable event predictions. In this paper, we introduce TimeCAP, a time-series processing framework that creatively employs Large Language Models (LLMs) as contextualizers of data, extending their typical usage predictors. TimeCAP incorporates two independent LLM agents:...
Large Language Models (LLMs) exhibit potential artificial generic intelligence recently, however, their usage is costly with high response latency. Given mixed LLMs own strengths and weaknesses, LLM routing aims to identify the most suitable model for each query in stream maximize quality minimize cost However, challenges involve: (1) dynamic trade-offs among quality, cost, latency; (2) enabling continual learning deployed systems; (3) navigating a varying (e.g., new addition or old removal)...
Graph neural network (GNN), as a powerful representation learning model on graph data, attracts much attention across various disciplines. However, recent studies show that GNN is vulnerable to adversarial attacks. How make more robust? What are the key vulnerabilities in GNN? address and defense against attacks? In this paper, we propose DefNet, an effective framework for GNNs. particular, first investigate latent every layer of GNNs corresponding strategies including dual-stage aggregation...