Donglin Wang

Dr. Donglin Wang

Machine Intelligence Lab (MiLAB), AI Division, School of Engineering, Westlake University, Zhejiang, China

Research Focus & Metrics

Academic Career Length

13Years (2014-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
11 April 2025 in Proceedings of the AAAI Conference on Artificial Intelligence
Verified
Journal Article
Open access
PDF file
HZHan Zhao
Han Zhao
MZMin Zhang
Min Zhang
WZWei Zhao
Wei Zhao
PDPengxiang Ding
Pengxiang Ding
SHSiteng Huang
Siteng Huang
Donglin WangDonglin Wang
Donglin Wang
In recent years, applying multi-modal large language models (MLLMs) in various fields has achieved remarkable success. However, as the foundation model for many downstream tasks, MLLMs comprise the well-known Transformer network, which has a less efficient quadratic computation complexity. In this study, we introduce Cobra, a multi-modal large-scale language model built upon a state-space model, which has demonstrated significant potential in efficiently handling long sequences with fast inference and linear scalability concerning sequence length. Specifically, Cobra involves replacing Transformer-based backbone models (e.g., LLaMA or Phi) with pre-trained Mamba language models. We then empirically explore effective strategies for aligning visual and textual modalities and integrating various pre-trained Mamba model variants with visual encoders. Experiments across various multi-modal benchmarks demonstrate that: (i) Cobra performs 3× ∼ 4× faster than the most computationally efficient state-of-the-art methods, e.g., LLaVA-Phi and MobileVLM v2. Additionally, its performance is significantly enhanced thanks to the implementation of linear sequential modeling. (ii) Cobra fine-tunes a small parameter (∼48% of model parameters), leading to a significant improvement in overall performance compared to LLaVA.

30 October 2024
Verified
Book Chapter
Closed access
PDF file
PDPengxiang Ding
Pengxiang Ding
HZHan Zhao
Han Zhao
WZWenjie Zhang
Wenjie Zhang
WSWenxuan Song
Wenxuan Song
MZMin Zhang
Min Zhang
Donglin WangDonglin Wang
Donglin Wang

14 August 2022
Verified
Conference Paper
Open access
YSYu Song
Yu Song
Donglin WangDonglin Wang
Donglin Wang
Graph Neural Networks (GNNs) are state-of-the-art models for performing prediction tasks on graphs. While existing GNNs have shown great performance on various tasks related to graphs, little attention has been paid to the scenario where out-of-distribution (OOD) nodes exist in the graph during training and inference. Borrowing the concept from CV and NLP, we define OOD nodes as nodes with labels unseen from the training set. Since a lot of networks are automatically constructed by programs, real-world graphs are often noisy and may contain nodes from unknown distributions. In this work, we define the problem of graph learning with out-of-distribution nodes. Specifically, we aim to accomplish two tasks: 1) detect nodes which do not belong to the known distribution and 2) classify the remaining nodes to be one of the known classes. We demonstrate that the connection patterns in graphs are informative for outlier detection, and propose Out-of-Distribution Graph Attention Network (OODGAT), a novel GNN model which explicitly models the interaction between different kinds of nodes and separate inliers from outliers during feature propagation. Extensive experiments show that OODGAT outperforms existing outlier detection methods by a large margin, while being better or comparable in terms of in-distribution classification.

15 November 2021 in IEEE Transactions on Neural Networks and Learning Systems
Verified
Journal Article
Closed access
TWTing Wang
Ting Wang
ZWZongkai Wu
Zongkai Wu
Donglin WangDonglin Wang
Donglin Wang
Vision-and-language navigation (VLN) is a challenging task that requires an agent to navigate in real-world environments by understanding natural language instructions and visual information received in real time. Prior works have implemented VLN tasks on continuous environments or physical robots, all of which use a fixed-camera configuration due to the limitations of datasets, such as 1.5-m height, 90° horizontal field of view (HFOV), and so on. However, real-life robots with different purposes have multiple camera configurations, and the huge gap in visual information makes it difficult to directly transfer the learned navigation skills between various robots. In this brief, we propose a visual perception generalization strategy based on meta-learning, which enables the agent to fast adapt to a new camera configuration. In the training phase, we first locate the generalization problem to the visual perception module and then compare two meta-learning algorithms for better generalization in seen and unseen environments. One of them uses the model-agnostic meta-learning (MAML) algorithm that requires few-shot adaptation, and the other refers to a metric-based meta-learning method with a feature-wise affine transformation (AT) layer. The experimental results on the VLN-CE dataset demonstrate that our strategy successfully adapts the learned navigation skills to new camera configurations, and the two algorithms show their advantages in seen and unseen environments respectively.


19 March 2018 in IEEE Systems Journal
Verified
Journal Article
Closed access
Donglin WangDonglin Wang
Donglin Wang
MFMichel Fattouche
Michel Fattouche
XZXingqun Zhan
Xingqun Zhan
This paper aims at investigating the feasibility of ranging and positioning in millimeter (mm)-level accuracy by adopting millimeter-wave (mmWave) and three-dimensional massive antenna arrays. The agent to be localized is equipped with massive uniform rectangular array (URA) and multiple anchors are considered to pursue a higher accuracy, where a far-field environment is assumed for phased massive URA. By considering both a static scenario where the agent with an unknown coordinate is stationary and a dynamic scenario where the agent is moving, the fundamental limits of time-based ranging and positioning are derived by Cr$\acute{\text {a}}$mer–Rao bound in a multipath model with and without a priori channel knowledge associated with mmWave transmission and massive URA. The relationship between the fundamental bound of range estimation and that of position estimation is theoretically clarified and the lower bound of position dilution of precision is derived. By investigating both models, the effect of the agent’s movement to ranging and positioning accuracy is observed. Numerical results show that the proposed localization network achieves a precise mm-level accuracy for ranging and positioning.