Junjie Huang

HiTIN: Hierarchy-aware Tree Isomorphism Network for Hierarchical Text Classification

2023

Hierarchical text classification (HTC) is a challenging subtask of multi-label classification as the labels form a complex hierarchical structure. Existing dual-encoder methods in HTC achieve weak performance gains with huge memory overheads and their structure …

Prism: Revealing Hidden Functional Clusters from Massive Instances in Cloud Systems

2023 · arXiv (Cornell University)

Ensuring the reliability of cloud systems is critical for both cloud vendors and customers. Cloud systems often rely on virtualization techniques to create instances of hardware resources, such as virtual machines. However, virtualization hinders the …

Unleashing the Potential of Multi-Channel Fusion in Retrieval for Personalized Recommendations

2024 · arXiv (Cornell University)

Recommender systems (RS) are pivotal in managing information overload in modern digital services. A key challenge in RS is efficiently processing vast item pools to deliver highly personalized recommendations under strict latency constraints. Multi-stage cascade …

SPA: A poisoning attack framework for graph neural networks through searching and pairing

2025 · Machine Learning

Graph Neural Networks (GNN) have played an important role in many fields, while GNNs also suffer from adversarial attacks that aim to malfunction the GNN model by changing the adjacency matrix (i.e. generating adversarial edges) …

Optimal and Almost Optimal Locally Repairable Codes from Hyperelliptic Curves

2025 · arXiv (Cornell University)

Locally repairable codes are widely applicable in contemporary large-scale distributed cloud storage systems and various other areas. By making use of some algebraic structures of elliptic curves, Li et al. developed a series of $q$-ary …

Learning to Compress Graphs via Dual Agents for Consistent Topological Robustness Evaluation

2026 · Proceedings of the AAAI Conference on Artificial Intelligence

As graph-structured data grow increasingly large, evaluating their robustness under adversarial attacks becomes computationally expensive and difficult to scale. To address this challenge, we propose to compress graphs into compact representations that preserve both topological …