Author name cluster

Hongkai Wen

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

11 papers

1 author row

IJCAI Conference 2025 Conference Paper

BGM: Demand Prediction for Expanding Bike-Sharing Systems with Dynamic Graph Modeling

Yixuan Zhao
Hongkai Wen
Xingchen Zhang
Man Luo

Accurate demand prediction is crucial for the equitable and sustainable expansion of bike-sharing systems, which help reduce urban congestion, promote low-carbon mobility, and improve transportation access in underserved areas. However, expanding these systems presents societal challenges, particularly in ensuring fair resource distribution and operational efficiency. A major hurdle is the difficulty of demand prediction at new stations, which lack historical usage data and are heavily influenced by the existing network. Additionally, new stations dynamically reshape demand patterns across time and space, complicating efforts to balance supply and accessibility in evolving urban environments. Existing methods model relationships between new and existing stations but often assume static patterns, overlooking how new stations reshape demand dynamics over time and space. To tackle these challenges, we propose a novel demand prediction framework for expanding bike-sharing systems, namely BGM, which leverages dynamic graph modeling to capture the evolving inter-station correlations while accounting for spatial and temporal heterogeneity. Specifically, we develop a knowledge transfer approach that studies the embeddings transformation across existing and new stations through a learnable orthogonal mapping matrix. We further design a gated selecting vector-based feature fusion mechanism to integrate the transferred embeddings and the intrinsic features of stations for precise predictions. Experiments on real-world bike-sharing data demonstrate that BGM outperforms existing methods.

PDF Details DOI

NeurIPS Conference 2025 Conference Paper

DCA: Graph-Guided Deep Embedding Clustering for Brain Atlases

Mo WANG
Kaining Peng
Jingsheng Tang
Hongkai Wen
Quanying Liu

Brain atlases are essential for reducing the dimensionality of neuroimaging data and enabling interpretable analysis. However, most existing atlases are predefined, group-level templates with limited flexibility and resolution. We present Deep Cluster Atlas (DCA), a graph-guided deep embedding clustering framework for generating individualized, voxel-wise brain parcellations. DCA combines a pretrained voxel-level fMRI autoencoder with spatially regularized deep clustering to produce functionally coherent and spatially contiguous regions. Our method supports flexible control over resolution and anatomical scope, and generalizes to arbitrary brain structures. We further introduce a standardized benchmarking platform for atlas evaluation, using multiple large-scale fMRI datasets. Across multiple datasets and scales, DCA outperforms state-of-the-art atlases, improving functional homogeneity by 98. 8% and silhouette coefficient by 29%, and achieves superior performance in downstream tasks. Furthermore, atlases demonstrate heterogeneous performance across various tasks and an atlas derived from a fine-tuned model yields superior results for its specific application. Codes are available at https: //github. com/ncclab-sustech/DCA.

PDF Details

IJCAI Conference 2025 Conference Paper

High-Fidelity Road Network Generation with Latent Diffusion Models

Jinming Wang
Hongkai Wen
Geyong Min
Man Luo

Road networks are the vein of modern cities. Yet, maintaining up-to-date and accurate road network information is a persistent challenge, especially in areas with rapid urban changes or limited surveying resources. Crowdsourced trajectories, e. g. , from GPS records collected by mobile devices and vehicles, have emerged as a powerful data source for continuously mapping the urban areas. However, the inherent noise, irregular and often sparse sampling rates, and the vast variability in movement patterns make the problem of road network generation from trajectories a non-trivial task. Existing methods often approach this from an appearance-based perspective: they typically render trajectories as 2D density maps and then employ heuristic algorithms to extract road networks - leading to inevitable information loss and thus poor performance especially when trajectories are sparse or ambiguities present, e. g. flyovers. In this paper, we propose a novel approach, called GraphWalker, to generate high-fidelity road network graphs from raw trajectories in an end-to-end manner. We achieve this by designing a bespoke latent diffusion transformer T2W-DiT, which treats input trajectories as generation conditions, and gradually denoises samples from a latent space to obtain the corresponding walks on the underlying road network graph - then assemble them together as the final road network. Extensive experiments on multiple datasets demonstrate the proposed GraphWalker can effectively generate high quality road networks from noisy and sparse trajectories, showcasing significant improvements over state-of-the-art.

PDF Details DOI

NeurIPS Conference 2025 Conference Paper

MixSignGraph: A Sign Sequence is Worth Mixed Graphs of Nodes

Shiwei Gan
Yafeng Yin
Zhiwei Jiang
Lei Xie
Sanglu Lu
Hongkai Wen

Recent advances in sign language research have benefited from CNN-based backbones, which are primarily transferred from traditional computer vision tasks (\eg object detection, image recognition). However, these CNN-based backbones usually excel at extracting features like contours and texture, but may struggle with capturing sign-related features. To capture such sign-related features, SignGraph model extracts the cross-region sign features by building the Local Sign Graph (LSG) module and the Temporal Sign Graph (TSG) module. However, we emphasize that although capturing cross-region dependencies can improve sign language performance, it may degrade the representation quality of local regions. To mitigate this, we introduce MixSignGraph, which represents sign sequences as a group of mixed graphs for feature extraction. Specifically, besides the LSG module and TSG module that model the intra-frame and inter-frame cross-regions features, we design a simple yet effective Hierarchical Sign Graph (HSG) module, which enhances local region representations following the extraction of cross-region features, by aggregating the same-region features from different-granularity feature maps of a frame, \ie to boost discriminative local features. In addition, to further improve the performance of gloss-free sign language task, we propose a simple yet counter-intuitive Text-based CTC Pre-training (TCTC) method, which generates pseudo gloss labels from text sequences for model pre-training. Extensive experiments conducted on the current five sign language datasets demonstrate that MixSignGraph surpasses the most current models on multiple sign language tasks across several datasets, without relying on any additional cues. Code and models are available at: \href{https: //github. com/gswycf/SignLanguage}{\textcolor{blue}{https: //github. com/gswycf/SignLanguage}}.

PDF Details

NeurIPS Conference 2025 Conference Paper

Shortcutting Pre-trained Flow Matching Diffusion Models is Almost Free Lunch

Xu Cai
Yang Wu
Qianli Chen
Haoran Wu
Lichuan Xiang
Hongkai Wen

We present an ultra-efficient post-training method for shortcutting large-scale pre-trained flow matching diffusion models into efficient few-step samplers, enabled by novel velocity field self-distillation. While shortcutting in flow matching, originally introduced by shortcut models, offers flexible trajectory-skipping capabilities, it requires a specialized step-size embedding incompatible with existing models unless retraining from scratch—a process nearly as costly as pretraining itself. Our key contribution is thus imparting a more aggressive shortcut mechanism to standard flow matching models (e. g. , Flux), leveraging a unique distillation principle that obviates the need for step-size embedding. Working on the velocity field rather than sample space and learning rapidly from self-guided distillation in an online manner, our approach trains efficiently, e. g. , producing a 3-step Flux <1 A100 day. Beyond distillation, our method can be incorporated into the pretraining stage itself, yielding models that inherently learn efficient, few-step flows without compromising quality. This capability also enables, to our knowledge, the first few-shot distillation method (e. g. , 10 text-image pairs) for dozen-billion-parameter diffusion models, delivering state-of-the-art performance at almost free cost.

PDF Details

NeurIPS Conference 2023 Conference Paper

EV-Eye: Rethinking High-frequency Eye Tracking through the Lenses of Event Cameras

Guangrong Zhao
Yurun Yang
Jingwei Liu
Ning Chen
Yiran Shen
Hongkai Wen
Guohao Lan

In this paper, we present EV-Eye, a first-of-its-kind large scale multimodal eye tracking dataset aimed at inspiring research on high-frequency eye/gaze tracking. EV-Eye utilizes an emerging bio-inspired event camera to capture independent pixel-level intensity changes induced by eye movements, achieving sub-microsecond latency. Our dataset was curated over a two-week period and collected from 48 participants encompassing diverse genders and age groups. It comprises over 1. 5 million near-eye grayscale images and 2. 7 billion event samples generated by two DAVIS346 event cameras. Additionally, the dataset contains 675 thousands scene images and 2. 7 million gaze references captured by Tobii Pro Glasses 3 eye tracker for cross-modality validation. Compared with existing event-based high-frequency eye tracking datasets, our dataset is significantly larger in size, and the gaze references involve more natural eye movement patterns, i. e. , fixation, saccade and smooth pursuit. Alongside the event data, we also present a hybrid eye tracking method as benchmark, which leverages both the near-eye grayscale images and event data for robust and high-frequency eye tracking. We show that our method achieves higher accuracy for both pupil and gaze estimation tasks compared to the existing solution.

PDF Details

NeurIPS Conference 2023 Conference Paper

Towards Data-Agnostic Pruning At Initialization: What Makes a Good Sparse Mask?

Hoang Pham
The Anh Ta
Shiwei Liu
Lichuan Xiang
Dung Le
Hongkai Wen
Long Tran-Thanh

Pruning at initialization (PaI) aims to remove weights of neural networks before training in pursuit of training efficiency besides the inference. While off-the-shelf PaI methods manage to find trainable subnetworks that outperform random pruning, their performance in terms of both accuracy and computational reduction is far from satisfactory compared to post-training pruning and the understanding of PaI is missing. For instance, recent studies show that existing PaI methods only able to find good layerwise sparsities not weights, as the discovered subnetworks are surprisingly resilient against layerwise random mask shuffling and weight re-initialization. In this paper, we study PaI from a brand-new perspective -- the topology of subnetworks. In particular, we propose a principled framework for analyzing the performance of Pruning and Initialization (PaI) methods with two quantities, namely, the number of effective paths and effective nodes. These quantities allow for a more comprehensive understanding of PaI methods, giving us an accurate assessment of different subnetworks at initialization. We systematically analyze the behavior of various PaI methods through our framework and observe a guiding principle for constructing effective subnetworks: *at a specific sparsity, the top-performing subnetwork always presents a good balance between the number of effective nodes and the number of effective paths. *Inspired by this observation, we present a novel data-agnostic pruning method by solving a multi-objective optimization problem. By conducting extensive experiments across different architectures and datasets, our results demonstrate that our approach outperforms state-of-the-art PaI methods while it is able to discover subnetworks that have much lower inference FLOPs (up to 3. 4$\times$). Code will be fully released.

PDF Details

AAAI Conference 2023 Conference Paper

Zero-Cost Operation Scoring in Differentiable Architecture Search

Lichuan Xiang
Lukasz Dudziak
Mohamed S. Abdelfattah
Thomas Chau
Nicholas D. Lane
Hongkai Wen

We formalize and analyze a fundamental component of dif- ferentiable neural architecture search (NAS): local “opera- tion scoring” at each operation choice. We view existing operation scoring functions as inexact proxies for accuracy, and we find that they perform poorly when analyzed empir- ically on NAS benchmarks. From this perspective, we intro- duce a novel perturbation-based zero-cost operation scor- ing (Zero-Cost-PT) approach, which utilizes zero-cost prox- ies that were recently studied in multi-trial NAS but de- grade significantly on larger search spaces, typical for dif- ferentiable NAS. We conduct a thorough empirical evalu- ation on a number of NAS benchmarks and large search spaces, from NAS-Bench-201, NAS-Bench-1Shot1, NAS- Bench-Macro, to DARTS-like and MobileNet-like spaces, showing significant improvements in both search time and accuracy. On the ImageNet classification task on the DARTS search space, our approach improved accuracy compared to the best current training-free methods (TE-NAS) while be- ing over 10× faster (total searching time 25 minutes on a single GPU), and observed significantly better transferabil- ity on architectures searched on the CIFAR-10 dataset with an accuracy increase of 1.8 pp. Our code is available at: https://github.com/zerocostptnas/zerocost operation score.

PDF Details DOI

NeurIPS Conference 2022 Conference Paper

BLOX: Macro Neural Architecture Search Benchmark and Algorithms

Thomas Chau
Łukasz Dudziak
Hongkai Wen
Nicholas Lane
Mohamed Abdelfattah

Neural architecture search (NAS) has been successfully used to design numerous high-performance neural networks. However, NAS is typically compute-intensive, so most existing approaches restrict the search to decide the operations and topological structure of a single block only, then the same block is stacked repeatedly to form an end-to-end model. Although such an approach reduces the size of search space, recent studies show that a macro search space, which allows blocks in a model to be different, can lead to better performance. To provide a systematic study of the performance of NAS algorithms on a macro search space, we release Blox – a benchmark that consists of 91k unique models trained on the CIFAR-100 dataset. The dataset also includes runtime measurements of all the models on a diverse set of hardware platforms. We perform extensive experiments to compare existing algorithms that are well studied on cell-based search spaces, with the emerging blockwise approaches that aim to make NAS scalable to much larger macro search spaces. The Blox benchmark and code are available at https: //github. com/SamsungLabs/blox.

PDF Details

IJCAI Conference 2020 Conference Paper

Rebalancing Expanding EV Sharing Systems with Deep Reinforcement Learning

Man Luo
Wenzhe Zhang
Tianyou Song
Kun Li
Hongming Zhu
Bowen Du
Hongkai Wen

Electric Vehicle (EV) sharing systems have recently experienced unprecedented growth across the world. One of the key challenges in their operation is vehicle rebalancing, i. e. , repositioning the EVs across stations to better satisfy future user demand. This is particularly challenging in the shared EV context, because i) the range of EVs is limited while charging time is substantial, which constrains the rebalancing options; and ii) as a new mobility trend, most of the current EV sharing systems are still continuously expanding their station networks, i. e. , the targets for rebalancing can change over time. To tackle these challenges, in this paper we model the rebalancing task as a Multi-Agent Reinforcement Learning (MARL) problem, which directly takes the range and charging properties of the EVs into account. We propose a novel approach of policy optimization with action cascading, which isolates the non-stationarity locally, and use two connected networks to solve the formulated MARL. We evaluate the proposed approach using a simulator calibrated with 1-year operation data from a real EV sharing system. Results show that our approach significantly outperforms the state-of-the-art, offering up to 14% gain in order satisfied rate and 12% increase in net revenue.

PDF Details DOI

AAAI Conference 2017 Conference Paper

VINet: Visual-Inertial Odometry as a Sequence-to-Sequence Learning Problem

Ronald Clark
Sen Wang
Hongkai Wen
Andrew Markham
Niki Trigoni

In this paper we present an on-manifold sequence-tosequence learning approach to motion estimation using visual and inertial sensors. It is to the best of our knowledge the ﬁrst end-to-end trainable method for visual-inertial odometry which performs fusion of the data at an intermediate feature-representation level. Our method has numerous advantages over traditional approaches. Speciﬁcally, it eliminates the need for tedious manual synchronization of the camera and IMU as well as eliminating the need for manual calibration between the IMU and camera. A further advantage is that our model naturally and elegantly incorporates domain speciﬁc information which signiﬁcantly mitigates drift. We show that our approach is competitive with state-of-theart traditional methods when accurate calibration data is available and can be trained to outperform them in the presence of calibration and synchronization errors.

PDF Details