Jiao Chen

Tao Zhang, Kehui Yao, Luyi Ma et al.

Evaluating large language models (LLMs) as judges is increasingly critical for building scalable and trustworthy evaluation pipelines. We present ScalingEval, a large-scale benchmarking study that systematically compares 36 LLMs, including GPT, Gemini, Claude, and Llama, across multiple product categories using a consensus-driven evaluation protocol. Our multi-agent framework aggregates pattern audits and issue codes into ground-truth labels via scalable majority voting, enabling reproducible comparison of LLM evaluators without human annotation. Applied to large-scale complementary-item recommendation, the benchmark reports four key findings: (i) Anthropic Claude 3.5 Sonnet achieves the highest decision confidence; (ii) Gemini 1.5 Pro offers the best overall performance across categories; (iii) GPT-4o provides the most favorable latency-accuracy-cost tradeoff; and (iv) GPT-OSS 20B leads among open-source models. Category-level analysis shows strong consensus in structured domains (Electronics, Sports) but persistent disagreement in lifestyle categories (Clothing, Food). These results establish ScalingEval as a reproducible benchmark and evaluation protocol for LLMs as judges, with actionable guidance on scaling, reliability, and model family tradeoffs.

Xiuding Cai, Xueyao Wang, Sen Wang et al.

Intraoperative monitoring and prediction of vital signs are critical for ensuring patient safety and improving surgical outcomes. Despite recent advances in deep learning models for medical time-series forecasting, several challenges persist, including the lack of standardized benchmarks, incomplete data, and limited cross-center validation. To address these challenges, we introduce VitalBench, a novel benchmark specifically designed for intraoperative vital sign prediction. VitalBench includes data from over 4,000 surgeries across two independent medical centers, offering three evaluation tracks: complete data, incomplete data, and cross-center generalization. This framework reflects the real-world complexities of clinical practice, minimizing reliance on extensive preprocessing and incorporating masked loss techniques for robust and unbiased model evaluation. By providing a standardized and unified platform for model development and comparison, VitalBench enables researchers to focus on architectural innovation while ensuring consistency in data handling. This work lays the foundation for advancing predictive models for intraoperative vital sign forecasting, ensuring that these models are not only accurate but also robust and adaptable across diverse clinical environments. Our code and data are available at https://github.com/XiudingCai/VitalBench.

Jiao Chen, Suyan Dai, Fangfang Chen et al.

Integrating large language models (LLMs) into autonomous driving enhances personalization and adaptability in open-world scenarios. However, traditional edge computing models still face significant challenges in processing complex driving data, particularly regarding real-time performance and system efficiency. To address these challenges, this study introduces EC-Drive, a novel edge-cloud collaborative autonomous driving system with data drift detection capabilities. EC-Drive utilizes drift detection algorithms to selectively upload critical data, including new obstacles and traffic pattern changes, to the cloud for processing by GPT-4, while routine data is efficiently managed by smaller LLMs on edge devices. This approach not only reduces inference latency but also improves system efficiency by optimizing communication resource use. Experimental validation confirms the system's robust processing capabilities and practical applicability in real-world driving conditions, demonstrating the effectiveness of this edge-cloud collaboration framework. Our data and system demonstration will be released at https://sites.google.com/view/ec-drive.

Xiuding Cai, Jiao Chen, Yaoyao Zhu et al.

Automated anesthesia promises to enable more precise and personalized anesthetic administration and free anesthesiologists from repetitive tasks, allowing them to focus on the most critical aspects of a patient's surgical care. Current research has typically focused on creating simulated environments from which agents can learn. These approaches have demonstrated good experimental results, but are still far from clinical application. In this paper, Policy Constraint Q-Learning (PCQL), a data-driven reinforcement learning algorithm for solving the problem of learning anesthesia strategies on real clinical datasets, is proposed. Conservative Q-Learning was first introduced to alleviate the problem of Q function overestimation in an offline context. A policy constraint term is added to agent training to keep the policy distribution of the agent and the anesthesiologist consistent to ensure safer decisions made by the agent in anesthesia scenarios. The effectiveness of PCQL was validated by extensive experiments on a real clinical anesthesia dataset. Experimental results show that PCQL is predicted to achieve higher gains than the baseline approach while maintaining good agreement with the reference dose given by the anesthesiologist, using less total dose, and being more responsive to the patient's vital signs. In addition, the confidence intervals of the agent were investigated, which were able to cover most of the clinical decisions of the anesthesiologist. Finally, an interpretable method, SHAP, was used to analyze the contributing components of the model predictions to increase the transparency of the model.

Jiao Chen, Jiayi He, Fangfang Chen et al.

Industrial AI is transitioning from traditional deep learning models to large-scale transformer-based architectures, with the Industrial Internet of Things (IIoT) playing a pivotal role. IIoT evolves from a simple data pipeline to an intelligent infrastructure, enabling and enhancing these advanced AI systems. This survey explores the integration of IIoT with large models (LMs) and their potential applications in industrial environments. We focus on four primary types of industrial LMs: language-based, vision-based, time-series, and multimodal models. The lifecycle of LMs is segmented into four critical phases: data foundation, model training, model connectivity, and continuous evolution. First, we analyze how IIoT provides abundant and diverse data resources, supporting the training and fine-tuning of LMs. Second, we discuss how IIoT offers an efficient training infrastructure in low-latency and bandwidth-optimized environments. Third, we highlight the deployment advantages of LMs within IIoT, emphasizing IIoT's role as a connectivity nexus fostering emergent intelligence through modular design, dynamic routing, and model merging to enhance system scalability and adaptability. Finally, we demonstrate how IIoT supports continual learning mechanisms, enabling LMs to adapt to dynamic industrial conditions and ensure long-term effectiveness. This paper underscores IIoT's critical role in the evolution of industrial intelligence with large models, offering a theoretical framework and actionable insights for future research.

Jiao Chen, Jianhua Tang, Xiaotong Yang et al.

AI coding agents demonstrate strong performance on general-purpose software benchmarks. However, their ability to handle 5G network engineering tasks remains unexplored. We propose SWE-Bench~5G, the first benchmark designed to investigate whether AI coding agents can resolve real-world bugs in 5G core network software. The benchmark collects task instances from three open-source 5G projects, packages each as a self-contained Docker environment with automated fail-to-pass tests, and provides a dual test strategy tailored to the complex runtime dependencies of telecom code. In addition, for instances whose original issues reference 3GPP specification clauses, we construct concise specification context documents, enabling controlled evaluation of whether domain knowledge improves agent performance. Experiments on four LLMs reveal that all models diagnose bugs at rates exceeding 91\%, yet resolve rates remain between 10\% and 30\%, suggesting that both iterative code editing capability and domain knowledge play important roles. The specification injection experiment further confirms that 3GPP excerpts improve resolve rates on specification-dependent bugs, while the gains on generic defensive checks remain limited, indicating that the effect of domain knowledge is conditional on bug type.

Jiao Chen, Jianhua Tang, Xiaotong Yang et al.

Autonomous 6G network management requires agents that can execute tools, observe the resulting state changes, and adapt their decisions accordingly. Existing benchmarks based on static questions or scripted episode replay, however, do not support such closed-loop interaction, limiting agents to passive evaluation without the ability to learn from environmental feedback. This paper presents 6GAgentGym to provide closed-loop capability. The framework provides an interactive environment with 42 typed tools whose effect classification distinguishes read-only observation from state-mutating configuration, backed by a learned Experiment Model calibrated on NS-3 simulation data. 6G-Forge bootstraps closed-loop training trajectories from NS-3 seeds via iterative Self-Instruct generation with execution verification against the Experiment Model. Supervised fine-tuning on the resulting corpus followed by reinforcement learning with online closed-loop interaction enables an 8B open-source model to achieve comparable overall success rate to GPT-5 on the accompanying 6GAgentBench, with stronger performance on long-horizon tasks. Together, these components provide a viable path toward autonomous, closed-loop network management.

Jiao Chen, Haoyi Wang, Jianhua Tang et al.

Low-altitude Unmanned Aerial Vehicle (UAV) networks rely on robust semantic segmentation as a foundational enabler for distributed sensing-communication-control co-design across heterogeneous agents within the network. However, segmentation foundation models deteriorate quickly under weather, lighting, and viewpoint drift. Resource-limited UAVs cannot run gradient-based test-time adaptation, while resource-massive UAVs adapt independently, wasting shared experience. To address these challenges, we propose AdaptFly, a prompt-guided test-time adaptation framework that adjusts segmentation models without weight updates. AdaptFly features two complementary adaptation modes. For resource-limited UAVs, it employs lightweight token-prompt retrieval from a shared global memory. For resource-massive UAVs, it uses gradient-free sparse visual prompt optimization via Covariance Matrix Adaptation Evolution Strategy. An activation-statistic detector triggers adaptation, while cross-UAV knowledge pool consolidates prompt knowledge and enables fleet-wide collaboration with negligible bandwidth overhead. Extensive experiments on UAVid and VDD benchmarks, along with real-world UAV deployments under diverse weather conditions, demonstrate that AdaptFly significantly improves segmentation accuracy and robustness over static models and state-of-the-art TTA baselines. The results highlight a practical path to resilient, communication-efficient perception in the emerging low-altitude economy.

Jiao Chen, Weihua Li, Jianhua Tang

In dynamic Industrial Internet of Things (IIoT) environments, models need the ability to selectively forget outdated or erroneous knowledge. However, existing methods typically rely on retain data to constrain model behavior, which increases computational and energy burdens and conflicts with industrial data silos and privacy compliance requirements. To address this, we propose a novel retain-free unlearning framework, referred to as Probing then Editing (PTE). PTE frames unlearning as a probe-edit process: first, it probes the decision boundary neighborhood of the model on the to-be-forgotten class via gradient ascent and generates corresponding editing instructions using the model's own predictions. Subsequently, a push-pull collaborative optimization is performed: the push branch actively dismantles the decision region of the target class using the editing instructions, while the pull branch applies masked knowledge distillation to anchor the model's knowledge on retained classes to their original states. Benefiting from this mechanism, PTE achieves efficient and balanced knowledge editing using only the to-be-forgotten data and the original model. Experimental results demonstrate that PTE achieves an excellent balance between unlearning effectiveness and model utility across multiple general and industrial benchmarks such as CWRU and SCUT-FD.

Xiang Zhang, Run He, Jiao Chen et al.

Class-incremental learning (CIL) enables models to learn new classes continually without forgetting previously acquired knowledge. Multi-label CIL (MLCIL) extends CIL to a real-world scenario where each sample may belong to multiple classes, introducing several challenges: label absence, which leads to incomplete historical information due to missing labels, and class imbalance, which results in the model bias toward majority classes. To address these challenges, we propose Label-Augmented Analytic Adaptation (L3A), an exemplar-free approach without storing past samples. L3A integrates two key modules. The pseudo-label (PL) module implements label augmentation by generating pseudo-labels for current phase samples, addressing the label absence problem. The weighted analytic classifier (WAC) derives a closed-form solution for neural networks. It introduces sample-specific weights to adaptively balance the class contribution and mitigate class imbalance. Experiments on MS-COCO and PASCAL VOC datasets demonstrate that L3A outperforms existing methods in MLCIL tasks. Our code is available at https://github.com/scut-zx/L3A.

Liulong Ma, Yanjie Liu, Jiao Chen et al.

Autonomous navigation is an essential capability of smart mobility for mobile robots. Traditional methods must have the environment map to plan a collision-free path in workspace. Deep reinforcement learning (DRL) is a promising technique to realize the autonomous navigation task without a map, with which deep neural network can fit the mapping from observation to reasonable action through explorations. It should not only memorize the trained target, but more importantly, the planner can reason out the unseen goal. We proposed a new motion planner based on deep reinforcement learning that can arrive at new targets that have not been trained before in the indoor environment with RGB image and odometry only. The model has a structure of stacked Long Short-Term memory (LSTM). Finally, experiments were implemented in both simulated and real environments. The source code is available: https://github.com/marooncn/navbot.

Liulong Ma, Yanjie Liu*, Jiao Chen

Robot navigation in mapless environment is one of the essential problems and challenges in mobile robots. Deep reinforcement learning is a promising technique to tackle the task of mapless navigation. Since reinforcement learning requires a lot of explorations, it is usually necessary to train the agent in the simulator and then migrate to the real environment. The big reality gap makes RGB image, the most common visual sensor, rarely used. In this paper we present a learning-based mapless motion planner by taking RGB images as visual inputs. Many parameters in end-to-end navigation network taking RGB images as visual input are used to extract visual features. Therefore, we decouple visual features extracted module from the reinforcement learning network to reduce the need of interactions between agent and environment. We use Variational Autoencoder (VAE) to encode the image, and input the obtained latent vector as low-dimensional visual features into the network together with the target and motion information, so that the sampling efficiency of the agent is greatly improved. We built simulation environment as robot navigation environment for algorithm comparison. In the test environment, the proposed method was compared with the end-to-end network, which proved its effectiveness and efficiency. The source code is available: https://github.com/marooncn/navbot.

Jiao Chen, Kehui Yao, Reza Yousefi Maragheh et al.

Current recommendation systems often require some form of textual data summarization, such as generating concise and coherent titles for product carousels or other grouped item displays. While large language models have shown promise in NLP domains for textual summarization, these approaches do not directly apply to recommendation systems, where explanations must be highly relevant to the core features of item sets, adhere to strict word limit constraints. In this paper, we propose CARTS (Collaborative Agents for Recommendation Textual Summarization), a multi-agent LLM framework designed for structured summarization in recommendation systems. CARTS decomposes the task into three stages-Generation Augmented Generation (GAG), refinement circle, and arbitration, where successive agent roles are responsible for extracting salient item features, iteratively refining candidate titles based on relevance and length feedback, and selecting the final title through a collaborative arbitration process. Experiments on large-scale e-commerce data and live A/B testing show that CARTS significantly outperforms single-pass and chain-of-thought LLM baselines, delivering higher title relevance and improved user engagement metrics.

Luyi Ma, Nikhil Thakurdesai, Jiao Chen et al.

Data processing is one of the fundamental steps in machine learning pipelines to ensure data quality. Majority of the applications consider the user-defined function (UDF) design pattern for data processing in databases. Although the UDF design pattern introduces flexibility, reusability and scalability, the increasing demand on machine learning pipelines brings three new challenges to this design pattern -- not low-code, not dependency-free and not knowledge-aware. To address these challenges, we propose a new design pattern that large language models (LLMs) could work as a generic data operator (LLM-GDO) for reliable data cleansing, transformation and modeling with their human-compatible performance. In the LLM-GDO design pattern, user-defined prompts (UDPs) are used to represent the data processing logic rather than implementations with a specific programming language. LLMs can be centrally maintained so users don't have to manage the dependencies at the run-time. Fine-tuning LLMs with domain-specific data could enhance the performance on the domain-specific tasks which makes data processing knowledge-aware. We illustrate these advantages with examples in different data processing tasks. Furthermore, we summarize the challenges and opportunities introduced by LLMs to provide a complete view of this design pattern for more discussions.

Jiao Chen, Jiayi He, Fangfang Chen et al.

Catastrophic forgetting remains a central challenge in continual learning (CL) with pre-trained models. While existing approaches typically freeze the backbone and fine-tune a small number of parameters to mitigate forgetting, they still rely on iterative error backpropagation and gradient-based optimization, which can be computationally intensive and less suitable for resource-constrained environments. To address this, we propose FoRo, a forward-only, gradient-free continual learning method. FoRo consists of a lightweight prompt tuning strategy and a novel knowledge encoding mechanism, both designed without modifying the pre-trained model. Specifically, prompt embeddings are inserted at the input layer and optimized using the Covariance Matrix Adaptation Evolution Strategy (CMA-ES), which mitigates distribution shifts and extracts high-quality task representations. Subsequently, task-specific knowledge is encoded into a knowledge encoding matrix via nonlinear random projection and recursive least squares, enabling incremental updates to the classifier without revisiting prior data. Experiments show that FoRo significantly reduces average forgetting and improves accuracy. Thanks to forward-only learning, FoRo reduces memory usage and run time while maintaining high knowledge retention across long task sequences. These results suggest that FoRo could serve as a promising direction for exploring continual learning with pre-trained models, especially in real-world multimedia applications where both efficiency and effectiveness are critical.

Jiayi He, Jiao Chen, Qianmiao Liu et al.

The Industrial Internet of Things (IIoT) integrates interconnected sensors and devices to support industrial applications, but its dynamic environments pose challenges related to data drift. Considering the limited resources and the need to effectively adapt models to new data distributions, this paper introduces a Continual Learning (CL) approach, i.e., Distillation-based Self-Guidance (DSG), to address challenges presented by industrial streaming data via a novel generative replay mechanism. DSG utilizes knowledge distillation to transfer knowledge from the previous diffusion-based generator to the updated one, improving both the stability of the generator and the quality of reproduced data, thereby enhancing the mitigation of catastrophic forgetting. Experimental results on CWRU, DSA, and WISDM datasets demonstrate the effectiveness of DSG. DSG outperforms the state-of-the-art baseline in accuracy, demonstrating improvements ranging from 2.9% to 5.0% on key datasets, showcasing its potential for practical industrial applications.

Jiao Chen, Luyi Ma, Xiaohan Li et al.

Knowledge Graphs (KGs) play a crucial role in enhancing e-commerce system performance by providing structured information about entities and their relationships, such as complementary or substitutable relations between products or product types, which can be utilized in recommender systems. However, relation labeling in KGs remains a challenging task due to the dynamic nature of e-commerce domains and the associated cost of human labor. Recently, breakthroughs in Large Language Models (LLMs) have shown surprising results in numerous natural language processing tasks. In this paper, we conduct an empirical study of LLMs for relation labeling in e-commerce KGs, investigating their powerful learning capabilities in natural language and effectiveness in predicting relations between product types with limited labeled data. We evaluate various LLMs, including PaLM and GPT-3.5, on benchmark datasets, demonstrating their ability to achieve competitive performance compared to humans on relation labeling tasks using just 1 to 5 labeled examples per relation. Additionally, we experiment with different prompt engineering techniques to examine their impact on model performance. Our results show that LLMs significantly outperform existing KG completion models in relation labeling for e-commerce KGs and exhibit performance strong enough to replace human labeling.