Licheng Zhang

Xin Wang, Liangtai Sun, Yaoming Zhu et al.

Existing code-generation benchmarks score a single mapping from a complete prompt to a one-shot output. However, real web development is different. Users seldom write a full spec at the start; many requirements only become clear once they look at an intermediate result and react to it. We present Asuka-Bench, a benchmark that pairs underspecified user intent with multi-round refinement, grounded in browser-rendered behavior. Each task is resolved through a closed loop: a Code Agent generates a web project, a UI Agent executes test cases on the deployed site, and a User LLM turns evaluation outcomes into natural-language feedback for the next round. The benchmark comprises 50 web tasks with 784 evaluation criteria and 2402 expected outcomes. We benchmark 8 LLMs across 2 agent frameworks. The results separate models clearly: weighted Task Pass Rate varies by 38 percentage points and models also differ substantially in their ability to repair from feedback. Asuka-Bench is also far from saturated: even the strongest model completes only 52% of projects after three rounds.

Shaohan Wang, Benfeng Xu, Licheng Zhang et al.

Deep Research Agents (DRAs) have demonstrated remarkable capabilities in autonomous information retrieval and report generation, showing great potential to assist humans in complex research tasks. Current evaluation frameworks primarily rely on LLM-generated references or LLM-derived evaluation dimensions. While these approaches offer scalability, they often lack the reliability of expert-verified content and struggle to provide objective, fine-grained assessments of critical dimensions. To bridge this gap, we introduce Wiki Live Challenge (WLC), a live benchmark that leverages the newest Wikipedia Good Articles (GAs) as expert-level references. Wikipedia's strict standards for neutrality, comprehensiveness, and verifiability serve as a great challenge for DRAs, with GAs representing the pinnacle of which. We curate a dataset of 100 recent Good Articles and propose Wiki Eval, a comprehensive evaluation framework comprising a fine-grained evaluation method with 39 criteria for writing quality and rigorous metrics for factual verifiability. Extensive experiments on various DRA systems demonstrate a significant gap between current DRAs and human expert-level Wikipedia articles, validating the effectiveness of WLC in advancing agent research. We release our benchmark at https://github.com/WangShao2000/Wiki_Live_Challenge

Jiaang Li, Quan Wang, Yi Liu et al.

Representation Learning on Knowledge Graphs (KGs) is essential for downstream tasks. The dominant approach, KG Embedding (KGE), represents entities with independent vectors and faces the scalability challenge. Recent studies propose an alternative way for parameter efficiency, which represents entities by composing entity-corresponding codewords matched from predefined small-scale codebooks. We refer to the process of obtaining corresponding codewords of each entity as entity quantization, for which previous works have designed complicated strategies. Surprisingly, this paper shows that simple random entity quantization can achieve similar results to current strategies. We analyze this phenomenon and reveal that entity codes, the quantization outcomes for expressing entities, have higher entropy at the code level and Jaccard distance at the codeword level under random entity quantization. Therefore, different entities become more easily distinguished, facilitating effective KG representation. The above results show that current quantization strategies are not critical for KG representation, and there is still room for improvement in entity distinguishability beyond current strategies. The code to reproduce our results is available at https://github.com/JiaangL/RandomQuantization.

Pengyu Wang, Benfeng Xu, Licheng Zhang et al.

Graph-based Retrieval-Augmented Generation (GraphRAG) organizes external knowledge as a hierarchical graph, enabling efficient retrieval and aggregation of scattered evidence across multiple documents. However, many existing benchmarks for GraphRAG rely on short, curated passages as external knowledge, failing to adequately evaluate systems in realistic settings involving long contexts and large-scale heterogeneous documents. To bridge this gap, we introduce WildGraphBench, a benchmark designed to assess GraphRAG performance in the wild. We leverage Wikipedia's unique structure, where cohesive narratives are grounded in long and heterogeneous external reference documents, to construct a benchmark reflecting real-word scenarios. Specifically, we sample articles across 12 top-level topics, using their external references as the retrieval corpus and citation-linked statements as ground truth, resulting in 1,100 questions spanning three levels of complexity: single-fact QA, multi-fact QA, and section-level summarization. Experiments across multiple baselines reveal that current GraphRAG pipelines help on multi-fact aggregation when evidence comes from a moderate number of sources, but this aggregation paradigm may overemphasize high-level statements at the expense of fine-grained details, leading to weaker performance on summarization tasks. Project page:https://github.com/BstWPY/WildGraphBench.

Licheng Zhang, Shuaili Liu, Qingsheng Yang et al.

In recent years, the rapid rise of video applications has led to an explosion of Internet video traffic, thereby posing severe challenges to network management. Therefore, effectively identifying and managing video traffic has become an urgent problem to be solved. However, the existing video traffic feature extraction methods mainly target at the traditional packet and flow level features, and the video traffic identification accuracy is low. Additionally, the issue of high data dimension often exists in video traffic identification, requiring an effective approach to select the most relevant features to complete the identification task. Although numerous studies have used feature selection to achieve improved identification performance, no feature selection research has focused on measuring feature distributions that do not overlap or have a small overlap. First, this study proposes to extract video-related features to construct a large-scale feature set to identify video traffic. Second, to reduce the cost of video traffic identification and select an effective feature subset, the current research proposes an adaptive distribution distance-based feature selection (ADDFS) method, which uses Wasserstein distance to measure the distance between feature distributions. To test the effectiveness of the proposal, we collected a set of video traffic from different platforms in a campus network environment and conducted a set of experiments using these data sets. Experimental results suggest that the proposed method can achieve high identification performance for video scene traffic and cloud game video traffic identification. Lastly, a comparison of ADDFS with other feature selection methods shows that ADDFS is a practical feature selection technique not only for video traffic identification, but also for general classification tasks.

Licheng Zhang, Bach Le, Naveed Akhtar et al.

Large Language Models (LLMs) have seen rapid advancements in recent years, with models like ChatGPT and DeepSeek, showcasing their remarkable capabilities across diverse domains. While substantial research has been conducted on LLMs in various fields, a comprehensive review focusing on their integration with Computer-Aided Design (CAD) remains notably absent. CAD is the industry standard for 3D modeling and plays a vital role in the design and development of products across different industries. As the complexity of modern designs increases, the potential for LLMs to enhance and streamline CAD workflows presents an exciting frontier. This article presents the first systematic survey exploring the intersection of LLMs and CAD. We begin by outlining the industrial significance of CAD, highlighting the need for AI-driven innovation. Next, we provide a detailed overview of the foundation of LLMs. We also examine both closed-source LLMs as well as publicly available models. The core of this review focuses on the various applications of LLMs in CAD, providing a taxonomy of six key areas where these models are making considerable impact. Finally, we propose several promising future directions for further advancements, which offer vast opportunities for innovation and are poised to shape the future of CAD technology. Github: https://github.com/lichengzhanguom/LLMs-CAD-Survey-Taxonomy

Jiahao Li, Quan Wang, Licheng Zhang et al.

In-context learning (ICL), which promotes inference with several demonstrations, has become a widespread paradigm to stimulate LLM capabilities for downstream tasks. Due to context length constraints, it cannot be further improved in spite of more training data, and general features directly from LLMs in ICL are not adaptive to the specific downstream task. In this paper, we propose a feature-adaptive and data-scalable in-context learning framework (FADS-ICL), which can leverage task-adaptive features to promote inference on the downstream task, with the supervision of beyond-context samples. Specifically, it first extracts general features of beyond-context samples via the LLM with ICL input form one by one, and introduces a task-specific modulator to perform feature refinement and prediction after fitting a specific downstream task. We conduct extensive experiments on FADS-ICL under varying data settings (4$\sim$128 shots) and LLM scale (0.8$\sim$70B) settings. Experimental results show that FADS-ICL consistently outperforms previous state-of-the-art methods by a significant margin under all settings, verifying the effectiveness and superiority of FADS-ICL. For example, under the 1.5B and 32 shots setting, FADS-ICL can achieve \textbf{+14.3} average accuracy from feature adaptation over vanilla ICL on 10 datasets, with \textbf{+6.2} average accuracy over the previous state-of-the-art method, and the performance can further improve with increasing training data. Code and data are publicly available at \url{https://github.com/jiahaozhenbang/FADS-ICL}.

Zihan Yang, Shuyuan Tu, Licheng Zhang et al.

Diffusion models have achieved remarkable generation quality, but they suffer from significant inference cost due to their reliance on multiple sequential denoising steps, motivating recent efforts to distill this inference process into a few-step regime. However, existing distillation methods typically approximate the teacher trajectory by using linear shortcuts, which makes it difficult to match its constantly changing tangent directions as velocities evolve across timesteps, thereby leading to quality degradation. To address this limitation, we propose ArcFlow, a few-step distillation framework that explicitly employs non-linear flow trajectories to approximate pre-trained teacher trajectories. Concretely, ArcFlow parameterizes the velocity field underlying the inference trajectory as a mixture of continuous momentum processes. This enables ArcFlow to capture velocity evolution and extrapolate coherent velocities to form a continuous non-linear trajectory within each denoising step. Importantly, this parameterization admits an analytical integration of this non-linear trajectory, which circumvents numerical discretization errors and results in high-precision approximation of the teacher trajectory. To train this parameterization into a few-step generator, we implement ArcFlow via trajectory distillation on pre-trained teacher models using lightweight adapters. This strategy ensures fast, stable convergence while preserving generative diversity and quality. Built on large-scale models (Qwen-Image-20B and FLUX.1-dev), ArcFlow only fine-tunes on less than 5% of original parameters and achieves a 40x speedup with 2 NFEs over the original multi-step teachers without significant quality degradation. Experiments on benchmarks show the effectiveness of ArcFlow both qualitatively and quantitatively.

Licheng Zhang, Bach Le, Naveed Akhtar et al.

Accurate detection and classification of diverse door types in floor plans drawings is critical for multiple applications, such as building compliance checking, and indoor scene understanding. Despite their importance, publicly available datasets specifically designed for fine-grained multi-class door detection remain scarce. In this work, we present a semi-automated pipeline that leverages a state-of-the-art object detector and a large language model (LLM) to construct a multi-class door detection dataset with minimal manual effort. Doors are first detected as a unified category using a deep object detection model. Next, an LLM classifies each detected instance based on its visual and contextual features. Finally, a human-in-the-loop stage ensures high-quality labels and bounding boxes. Our method significantly reduces annotation cost while producing a dataset suitable for benchmarking neural models in floor plan analysis. This work demonstrates the potential of combining deep learning and multimodal reasoning for efficient dataset construction in complex real-world domains.

Licheng Zhang, Bach Le, Naveed Akhtar et al.

Building compliance checking (BCC) is a critical process for ensuring that constructed facilities meet regulatory standards. A core component of BCC is the accurate enumeration of facility types and their spatial distribution. Despite its importance, this problem has been largely overlooked in the literature, posing a significant challenge for BCC and leaving a critical gap in existing workflows. Performing this task manually is time-consuming and labor-intensive. Recent advances in large language models (LLMs) offer new opportunities to enhance automation by combining visual recognition with reasoning capabilities. In this paper, we introduce a new task for BCC: automated facility enumeration, which involves validating the quantity of each facility type against statutory requirements. To address it, we propose a novel method that integrates door detection with LLM-based reasoning. We are the first to apply LLMs to this task and further enhance their performance through a Chain-of-Thought (CoT) pipeline. Our approach generalizes well across diverse datasets and facility types. Experiments on both real-world and synthetic floor plan data demonstrate the effectiveness and robustness of our method.

Liheng Wang, Licheng Zhang, Hailin Xu et al.

For bone segmentation, the classical geodesic active contour model is usually limited by its indiscriminate feature extraction, and then struggles to handle the phenomena of edge obstruction, edge leakage and bone fracture. Thus, we propose a fracture interactive geodesic active contour algorithm tailored for bone segmentation, which can better capture bone features and perform robustly to the presence of bone fractures and soft tissues. Inspired by orthopedic knowledge, we construct a novel edge-detector function that combines the intensity and gradient norm, which guides the contour towards bone edges without being obstructed by other soft tissues and therefore reduces mis-segmentation. Furthermore, distance information, where fracture prompts can be embedded, is introduced into the contour evolution as an adaptive step size to stabilize the evolution and help the contour stop at bone edges and fractures. This embedding provides a way to interact with bone fractures and improves the accuracy in the fracture regions. Experiments in pelvic and ankle segmentation demonstrate the effectiveness on addressing the aforementioned problems and show an accurate, stable and consistent performance, indicating a broader application in other bone anatomies. Our algorithm also provides insights into combining the domain knowledge and deep neural networks.

Shaohan Wang, Licheng Zhang, Zheren Fu et al.

Retrieval-Augmented Generation (RAG) is an effective method to enhance the capabilities of large language models (LLMs). Existing methods typically optimize the retriever or the generator in a RAG system by directly using the top-k retrieved documents. However, two key issues inherent in the training data constrain the effectiveness of this training paradigm: (1) across different queries, the top-k retrieved documents vary greatly in content quality, with some providing valuable knowledge while others lack critical information or are even misleading, and training on such data in a purely random manner may impair the generator's ability to extract key information; (2) for a given query, the limited set of k documents often exhibits low discriminability, and training solely on them makes it difficult for the retriever to learn how to distinguish between relevant and irrelevant documents. To address these issues, we introduce DACL-RAG, a multi-stage RAG training framework that combines a multi-level Data Augmentation strategy with a multi-stage Curriculum Learning paradigm. The data augmentation strategy constructs comprehensive and diverse training sets with controllable difficulty levels through sample evolution, while the curriculum learning paradigm organizes them into progressive stages for training, ensuring stable and consistent improvements, thereby optimizing the overall performance and generalization of the RAG system more effectively. Our DACL-RAG framework demonstrates consistent effectiveness across four open-domain QA datasets, achieving performance gains of 2% to 4% over multiple advanced methods.

Cheng Zhan, Licheng Zhang, Xin Zhao et al.

Over the year, people have been using deep learning to tackle inversion problems, and we see the framework has been applied to build relationship between recording wavefield and velocity (Yang et al., 2016). Here we will extend the work from 2 perspectives, one is deriving a more appropriate loss function, as we now, pixel-2-pixel comparison might not be the best choice to characterize image structure, and we will elaborate on how to construct cost function to capture high level feature to enhance the model performance. Another dimension is searching for the more appropriate neural architecture, which is a subset of an even bigger picture, the automatic machine learning, or AutoML. There are several famous networks, U-net, ResNet (He et al., 2016) and DenseNet (Huang et al., 2017), and they achieve phenomenal results for certain problems, yet it's hard to argue they are the best for inversion problems without thoroughly searching within certain space. Here we will be showing our architecture search results for inversion.

Cheng Zhan, Licheng Zhang, Zhenzhen Zhong et al.

Deep Learning is gaining traction with geophysics community to understand subsurface structures, such as fault detection or salt body in seismic data. This study describes using deep learning method for iceberg or ship recognition with synthetic aperture radar (SAR) data. Drifting icebergs pose a potential threat to activities offshore around the Arctic, including for both ship navigation and oil rigs. Advancement of satellite imagery using weather-independent cross-polarized radar has enabled us to monitor and delineate icebergs and ships, however a human component is needed to classify the images. Here we present Transfer Learning, a convolutional neural network (CNN) designed to work with a limited training data and features, while demonstrating its effectiveness in this problem. Key aspect of the approach is data augmentation and stacking of multiple outputs, resulted in a significant boost in accuracy (logarithmic score of 0.1463). This algorithm has been tested through participation at the Statoil/C-Core Kaggle competition.

Zhenzhen Zhong, Shujiao Huang, Cheng Zhan et al.

Large-scale datasets have played a significant role in progress of neural network and deep learning areas. YouTube-8M is such a benchmark dataset for general multi-label video classification. It was created from over 7 million YouTube videos (450,000 hours of video) and includes video labels from a vocabulary of 4716 classes (3.4 labels/video on average). It also comes with pre-extracted audio & visual features from every second of video (3.2 billion feature vectors in total). Google cloud recently released the datasets and organized 'Google Cloud & YouTube-8M Video Understanding Challenge' on Kaggle. Competitors are challenged to develop classification algorithms that assign video-level labels using the new and improved Youtube-8M V2 dataset. Inspired by the competition, we started exploration of audio understanding and classification using deep learning algorithms and ensemble methods. We built several baseline predictions according to the benchmark paper and public github tensorflow code. Furthermore, we improved global prediction accuracy (GAP) from base level 77% to 80.7% through approaches of ensemble.