Jian Li

Reform and Practice of the Trinity Teaching Mode for the Course Mechanical Control Engineering Base

2016

Traditional training program, assessment requirements and evaluation criteria can already not meet the requirements of college students in the new era because of the characteristics of the Course Mechanical Engineering Control Base, the course is …

Experimental realization of a 2 × 2 polarization-independent split-ratio-tunable optical beam splitter

2016 · Optics Express

We realized a polarization-independent split-ratio-tunable optical beam splitter supporting two input and output ports through a stable interferometer. By adjusting the angle of a half-wave plate in the interferometer, we can tune the beam splitter …

Radar-Vision Fusion for Correcting the Position of Target Vehicles

2018

Calculation of the exact coordinate of the target position is a crucial problem in the field of vehicle detection. This paper describes a vehicle detection system fusing radar and vision data which improve the accuracy …

LORI: A Learning-to-Rank-Based Integration Method of Location Recommendation

2019 · IEEE Transactions on Computational Social Systems

Location recommendation method is an important application in a location-based social network. At present, it is a trend to integrate different recommendation methods since they have their own advantages in capturing different preferences of users …

Towards Instance Optimal Bounds for Best Arm Identification

2016 · Conference on Learning Theory

In the classical best arm identification (Best-$1$-Arm) problem, we are given $n$ stochastic bandit arms, each associated with a reward distribution with an unknown mean. We would like to identify the arm with the largest …

Information Aggregation for Multi-Head Attention with Routing-by-Agreement

2019

Jian Li, Baosong Yang, Zi-Yi Dou, Xing Wang, Michael R. Lyu, Zhaopeng Tu. Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long …

Learning Vector-valued Functions with Local Rademacher Complexity and Unlabeled Data

2019 · arXiv (Cornell University)

We consider a general family of problems of which the output space admits vector-valued structure, covering a broad family of important domains, e.g. multi-label learning and multi-class classification. By using local Rademacher complexity and unlabeled …

Distributed privacy preserving technology in dynamic networks

2019 · International Journal of High Performance Computing and Networking

With the development of information technology, large-scale social network graph data have been produced and released to provide data analysis for scientific research and business structures, while traditional network privacy protection technology does not meet …

High-fidelity, high-scalability two-qubit gate scheme for superconducting qubits

2020 · arXiv (Cornell University)

High-quality two-qubit gate operations are crucial for scalable quantum information processing. Often, the gate fidelity is compromised when the system becomes more integrated. Therefore, a low-error-rate, easy-to-scale two-qubit gate scheme is highly desirable. Here, we …

Suppressing Coherent Two-Qubit Errors via Dynamical Decoupling

2021 · Physical Review Applied

Scalable quantum information processing requires the ability to tune multiqubit interactions. This makes the precise manipulation of quantum states particularly difficult for multiqubit interactions because tunability unavoidably introduces sensitivity to fluctuations in the tuning parameters, …

State Evaluation of Electronic Transformers Based on Cross Weight Method AHP

2021

This paper focuses on the problems of frequent faults, abnormal events, and the lack of operational state evaluation means for electronic transformers. Based on a large amount of equipment operation and state data, a cross …

Straggler-Resilient Decentralized Learning via Adaptive Asynchronous Updates

2023 · arXiv (Cornell University)

With the increasing demand for large-scale training of machine learning models, fully decentralized optimization methods have recently been advocated as alternatives to the popular parameter server framework. In this paradigm, each worker maintains a local …

Efficient semi-quantum secret sharing protocol using single particles

2023 · Chinese Physics B

Semi-quantum secret sharing (SQSS) is a branch of quantum cryptography which only requires the dealer to have quantum capabilities, reducing the difficulty of protocol implementation. However, the efficiency of the SQSS protocol still needs to …

AKE-GNN: Effective Graph Learning with Adaptive Knowledge Exchange

2023

Graph Neural Networks (GNNs) have already been widely used in various graph mining tasks. However, recent works reveal that the learned weights (channels) in well-trained GNNs are highly redundant, which inevitably limits the performance of …

Adversarial Preference Optimization: Enhancing Your Alignment via RM-LLM Game

2023 · arXiv (Cornell University)

Human preference alignment is essential to improve the interaction quality of large language models (LLMs). Existing alignment methods depend on manually annotated preference data to guide the LLM optimization directions. However, continuously updating LLMs for …

FedNS: A Fast Sketching Newton-Type Algorithm for Federated Learning

2024 · Proceedings of the AAAI Conference on Artificial Intelligence

Recent Newton-type federated learning algorithms have demonstrated linear convergence with respect to the communication rounds. However, communicating Hessian matrices is often unfeasible due to their quadratic communication complexity. In this paper, we introduce a novel …

Key-Point-Driven Mathematical Reasoning Distillation of Large Language Model

2024 · arXiv (Cornell University)

Large Language Models (LLMs) have demonstrated exceptional proficiency in mathematical reasoning tasks due to their extensive parameter counts and training on vast datasets. Despite these capabilities, deploying LLMs is hindered by their computational demands. Distilling …

Clustering-Enhanced Multimodal Pre-Training for Histology-Gene Joint Representation Learning

2025

Computational pathology has emerged as a powerful tool for developing prognostic models from histology images. Recent advances in multimodal approaches have demonstrated that integrating whole-slide images (WSIs) with bulk transcriptomics data enhances patient outcome predictions …

AIMMerging: Adaptive Iterative Model Merging Using Training Trajectories for Language Model Continual Learning

2025 · arXiv (Cornell University)

Continual learning (CL) is essential for deploying large language models (LLMs) in dynamic real-world environments without the need for costly retraining. Recent model merging-based methods have attracted significant attention, but they still struggle to effectively …

NetSMF: Large-Scale Network Embedding as Sparse Matrix Factorization

2019

We study the problem of large-scale network embedding, which aims to learn latent representations for network mining applications. Previous research shows that 1) popular network embedding benchmarks, such as DeepWalk, are in essence implicitly factorizing …

Multi-Head Attention with Disagreement Regularization

2018

Multi-head attention is appealing for the ability to jointly attend to information from different representation subspaces at different positions. In this work, we introduce a disagreement regularization to explicitly encourage the diversity among multiple attention …