Author name cluster

Zhicheng Wang

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.

7 papers

2 author rows

AAAI Conference 2026 Conference Paper

IPFormer: Instance Prompt-guided Transformer for Multi-modal Multi-shot Video Understanding

Yujia Liang
Jile Jiao
Xuetao Feng
Xinchen Liu
Kun Liu
Yuan Wang
Zixuan Ye
Hao Lu

Video Large Language Models (VideoLLMs), which adopt large language models for video understanding, have been demonstrated for single-shot videos. However, they usually struggle in multi-shot videos with frequent shot changes, varying camera angles, etc., which makes VideoLLMs hardly answer questions about multiple instances or shots over the whole video. We attribute this challenge to two issues: 1) the lack of multi-shot multi-instance annotations of existing datasets, and 2) the negligence of instance-aware modeling of current VideoLLMs. Therefore, we first introduce a new dataset termed MultiClip-Bench, featuring dense descriptions and question-answering pairs tailored for multi-shot and multi-instance scenarios. Moreover, since the existing VideoLLMs neglect the explicit modeling of instance-related features, we propose a novel Instance Prompt-guided Transformer, named IPFormer, to achieve instance-aware videounderstanding. In the IPFormer, we design a simple but effective instance-aware feature injection module, which encodes instance features as instance prompts via an attention-based connector. By this means, IPFormer can aggregate instance-specific information across multiple shots. Extensive experiments not only show that our dataset and model significantly improve multi-shot video understanding. but also show that our MultiClip-Bench can provide valuable training data and benchmarks for various video understanding tasks.

PDF Details DOI

EAAI Journal 2025 Journal Article

A thermodynamics-informed deep learning approach for lightweight modeling of gas turbine performance

Xiaomo Jiang
Yiyang Liu
Manman Wei
Xueyu Cheng
Zhicheng Wang

Lightweight performance modeling approaches are particularly crucial in the condition monitoring system of a heavy-duty gas turbine (HDGT) to track its efficiency changes over time accurately and promptly. This ensures sufficient productivity through a well-planned predictive maintenance strategy. This paper introduces a physics-informed deep learning methodology for lightweight modeling of HDGT performance, aimed at a real-time degradation monitoring for predictive maintenance. Initially, a mechanism-based thermodynamic model is established to serve as a performance benchmark. Subsequently, corrections are applied by adjusting base-load operating conditions to reference conditions, thereby mitigating the influence of ambient conditions and power demand. A substitute model called deep operator networks (DeepONet) is then constructed by integrating actual and simulation data to rapidly obtain crucial gas turbine performance parameters, such as compressor and turbine efficiency, as well as corrected power output and heat rate of the system. A standardized procedure is developed to automate the efficient performance modeling for gas turbines. To showcase the benefits of the proposed methodology and procedure, a comparative study with three different classical models is conducted using data from two various real-world HDGT machines. The DeepONet deep learning model is utilized to rapidly generate multivariate performance results for a gas turbine at 10ms for a single-step prediction, significantly faster than the physics-based model, which takes 6s. The prediction error is less than 0. 1% on average when compared to the latter. Numerical results demonstrate that the proposed methodology offers a promising tool for real-time performance prediction of a gas turbine for predictive maintenance purposes.

Details DOI

IROS Conference 2025 Conference Paper

Constrained Behavior Cloning for Robotic Learning

Jun Xie
Jianwei Tan
Wensheng Liang
Zhicheng Wang
Xiaoguang Ma

Behavior cloning (BC) is a widely used method for learning from expert demonstrations due to its simplicity and efficiency. However, the reliability and stability of BC decline when facing data distribution shifts, especially in single-arm robots with limited fields of view. This study introduces a Geometrically and Historically Constrained Behavior Cloning (GHCBC) method, where an HCBC module utilizes visual and action histories to capture temporal dependencies, maximizing the use of available information, and a GCBC module incorporates high-level perceptual data, such as the relative poses of joints and end-effectors, to enhance BC performance. Experiments demonstrate that the GHCBC outperforms current SOTA BC methods, achieving a 31. 5% improvement in simulation success rates and 48. 4% in real-robot scenarios respectively. To the best of our knowledge, this is the first time that the GHCBC has been introduced in robotic BC where great potential is demonstrated for long-term tasks in real world environments.

Details

ECAI Conference 2025 Conference Paper

Subconscious Robotic Imitation Learning

Jun Xie
Zhicheng Wang
Jianwei Tan
Huanxu Lin
Yang Jiang
Xiaoguang Ma

While imitation learning (IL) emerges as a promising paradigm for embodied intelligent robots, its practical application is constrained by slow execution speeds, caused by the computational intensity of precise multi-model trajectory prediction, especially in complex dynamic environments. In contrast, humans can efficiently perform long-duration tasks through subconscious-driven habitual actions, such as riding bikes, without focusing on execution details. Motivated by this insight, we proposed Subconscious Robotic Imitation Learning (SRIL) framework, which mimicked the subconscious information extraction and decision-making abilities through intent-aware sampling and cognitive hierarchical reasoning, thereby significantly improving IL task execution efficiency. Experimental results demonstrated that execution speeds of the SRIL were 100% to 200% faster over SOTA policies for comprehensive bimanual tasks, with consistently higher success rates.

Details

AAAI Conference 2024 Conference Paper

Vision Transformer Off-the-Shelf: A Surprising Baseline for Few-Shot Class-Agnostic Counting

Zhicheng Wang
Liwen Xiao
Zhiguo Cao
Hao Lu

Class-agnostic counting (CAC) aims to count objects of interest from a query image given few exemplars. This task is typically addressed by extracting the features of query image and exemplars respectively and then matching their feature similarity, leading to an extract-then-match paradigm. In this work, we show that CAC can be simplified in an extract-and-match manner, particularly using a vision transformer (ViT) where feature extraction and similarity matching are executed simultaneously within the self-attention. We reveal the rationale of such simplification from a decoupled view of the self-attention.The resulting model, termed CACViT, simplifies the CAC pipeline into a single pretrained plain ViT. Further, to compensate the loss of the scale and the order-of-magnitude information due to resizing and normalization in plain ViT, we present two effective strategies for scale and magnitude embedding. Extensive experiments on the FSC147 and the CARPK datasets show that CACViT significantly outperforms state-of-the-art CAC approaches in both effectiveness (23.60% error reduction) and generalization, which suggests CACViT provides a concise and strong baseline for CAC. Code will be available.

PDF Details DOI

IJCAI Conference 2023 Conference Paper

Reconstruction-Aware Prior Distillation for Semi-supervised Point Cloud Completion

Zhaoxin Fan
Yulin He
Zhicheng Wang
Kejian Wu
Hongyan Liu
Jun He

Real-world sensors often produce incomplete, irregular, and noisy point clouds, making point cloud completion increasingly important. However, most existing completion methods rely on large paired datasets for training, which is labor-intensive. This paper proposes RaPD, a novel semi-supervised point cloud completion method that reduces the need for paired datasets. RaPD utilizes a two-stage training scheme, where a deep semantic prior is learned in stage 1 from unpaired complete and incomplete point clouds, and a semi-supervised prior distillation process is introduced in stage 2 to train a completion network using only a small number of paired samples. Additionally, a self-supervised completion module is introduced to improve performance using unpaired incomplete point clouds. Experiments on multiple datasets show that RaPD outperforms previous methods in both homologous and heterologous scenarios.

PDF Details DOI

AAAI Conference 2021 Conference Paper

DeepTrader: A Deep Reinforcement Learning Approach for Risk-Return Balanced Portfolio Management with Market Conditions Embedding

Zhicheng Wang
Biwei Huang
Shikui Tu
Kun Zhang
Lei Xu

Most existing reinforcement learning (RL)-based portfolio management models do not take into account the market conditions, which limits their performance in risk-return balancing. In this paper, we propose Deep- Trader, a deep RL method to optimize the investment policy. In particular, to tackle the risk-return balancing problem, our model embeds macro market conditions as an indicator to dynamically adjust the proportion between long and short funds, to lower the risk of market fluctuations, with the negative maximum drawdown as the reward function. Additionally, the model involves a unit to evaluate individual assets, which learns dynamic patterns from historical data with the price rising rate as the reward function. Both temporal and spatial dependencies between assets are captured hierarchically by a specific type of graph structure. Particularly, we find that the estimated causal structure best captures the interrelationships between assets, compared to industry classification and correlation. The two units are complementary and integrated to generate a suitable portfolio which fits the market trend well and strikes a balance between return and risk effectively. Experiments on three well-known stock indexes demonstrate the superiority of DeepTrader in terms of risk-gain criteria.

PDF Details