KGKaifeng Guo

Kaifeng Guo

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Kaifeng Guo has a total of 19 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Hierarchical Residual Policy Optimization for Generative Recommendations

    by Kaifeng Guo, Yiming Yang, Jingtong Gao, Guolei Zeng, Fukang Yang, Yukang Liang, Peng Jiang, Qingpeng Cai, Xiangyu Zhao

    Generative recommenders select items by autoregressively decoding semantic identifiers (SIDs), whose token positions induce a coarse-to-fine hierarchy over the item space. In practice, SID decoders are trained via supervised next-token prediction, which imitates logged trajectories rather than directly optimizing downstream utility. This motivates post-training with outcome feedback to guide decoding toward higher utility. However, logged feedback is only observed for the final exposed item, causing most post-training methods to operate at the item level and broadcast the same terminal signal across all SID tokens. As a result, token-level credit assignment becomes sparse, high-variance, and layer-dependent. To this end, we propose Hierarchical Residual Policy Optimization (HRPO), a post-training framework that converts item-level outcomes into dense, token-aligned learning signals for conservative token-wise improvement. Specifically, HRPO first estimates SID prefix-level utilities via group-wise reward smoothing over feature-based user clusters. It then decomposes these utilities into residual token credits and accumulates them into credit-to-go signals. Finally, Residual-Return Policy Optimization (RRPO) optimizes the residual credits using clipped updates, group-normalized advantages, and KL regularization to preserve stability. Experiments on a public dataset and an online A/B test in a large-scale commercial system show consistent gains in session-level utility and key business metrics. Source code and the archived artifact are available for reproduction.

  • Open Access

    Deep Leakage From Horizontal Federated Sequential Recommender Systems

    by Kaifeng Guo, Kesheng Xie, Zian Shi, Rongjian Gao

    Shared gradients are extensively utilized for safeguarding the privacy of training data. However, an increasing body of research is uncovering that the gradients or model parameters transmitted in distributed systems may also leak users’ private information. A majority of these studies are predicated on intercepting the data transmitted by individual users to derive their private information. A smaller portion of research can extrapolate a multitude of users’ private data through the average gradients passed by the server. These investigations have analyzed information leakage in image and text domains, yet have not explored the leakage issues inherent within recommendation systems operating in distributed environments. Furthermore, the impact of batch size and model parameters on the degree of leakage has not been sufficiently analyzed. This paper proposes that within recommendation systems, it is feasible to infer the encapsulated user privacy data by receiving average gradients or model parameters provided by the server. Additionally, the paper evaluates the extent to which various parameters within the system impact the leakage and presents corresponding defensive measures while assessing their efficacy.


  • Open Access

    Deep learning in finance assessing twitter sentiment impact and prediction on stocks

    by Kaifeng Guo, Haoling Xie

    The widespread adoption of social media platforms has led to an influx of data that reflects public sentiment, presenting a novel opportunity for market analysis. This research aims to quantify the correlation between the fleeting sentiments expressed on social media and the measurable fluctuations in the stock market. By adapting a pre-existing sentiment analysis algorithm, we refined a model specifically for evaluating the sentiment of tweets associated with financial markets. The model was trained and validated against a comprehensive dataset of stock-related discussions on Twitter, allowing for the identification of subtle emotional cues that may predict changes in stock prices. Our quantitative approach and methodical testing have revealed a statistically significant relationship between sentiment expressed on Twitter and subsequent stock market activity. These findings suggest that machine learning algorithms can be instrumental in enhancing the analytical capabilities of financial experts. This article details the technical methodologies used, the obstacles overcome, and the potential benefits of integrating machine learning-based sentiment analysis into the realm of economic forecasting.


  • Open Access

    Graph convolutional network and self-attentive for sequential recommendation

    by Kaifeng Guo, Guolei Zeng

    Sequential recommender systems (SRS) aim to provide personalized recommendations to users in the context of large-scale datasets and complex user behavior sequences. However, the effectiveness of most existing embedding techniques in capturing the intricate relationships between items remains suboptimal, with a significant concentration of item embedding vectors that hinder the improvement of final prediction performance. Nevertheless, our study reveals that the distribution of item embeddings can be effectively dispersed through graph interaction networks and contrastive learning. In this article, we propose a graph convolutional neural network to capture the complex relationships between users and items, leveraging the learned embedding vectors of nodes to represent items. Additionally, we employ a self-attentive sequential model to predict outcomes based on the item embedding sequences of individual users. Furthermore, we incorporate instance-wise contrastive learning (ICL) and prototype contrastive learning (PCL) during the training process to enhance the effectiveness of representation learning. Broad comparative experiments and ablation studies were conducted across four distinct datasets. The experimental outcomes clearly demonstrate the superior performance of our proposed GSASRec model.


  • Open Access

    Multi-Task Learning for Lung Disease Classification and Report Generation via Prior Graph Structure and Contrastive Learning

    by Kaifeng Guo, Shihao Zheng, Ri Huang, Rongjian Gao

    Analyzing medical records takes a large amount of time and resources for medical workers. Therefore, the advancement of report generation technology is critical for saving doctors’ time while producing reports. Some diseases frequently exhibit co-occurrence interactions in the context of medical picture categorization and report creation, which necessitate hu-man-designed and added representation. To further capture this latent information, we propose creating a disease co-occurrence matrix from the existing dataset and training a graph neural network on it. We use multi-label contrastive learning to assist the model in distinguishing the mutual links between different diseases. Meanwhile, we incorporate report retrieval module and decoder model to complete the report generation task, and the multi-task learning of disease classification and report generation can improve the generalization ability of the target task. Experimental results show that our pro-posed combined method and multi-task learning approach have shown significant improvement compared to previous research.


  • Open Access

    A Joint Radar Signal Sorting Method For Multi-Radar Reconnaissance Station

    by He Kun, Liu Xubo, Liu Yi, GUO RiCheng, GUO KaiFeng

    Passive Detection Network Composed of Multi-Radar Reconnaissance Station, which can receive multiple sets of radar signal pulse descriptor word at the same time, joint signal sorting becomes an inevitable measure of such data. This paper proposes a multi-radar time-aligned association sorting method, and builds scenario close to actual conditions for testing. The results show that method can overcome the problem of partial pulse loss leading to sorting failure, and can quickly complete complex radar signal sorting in real electromagnetic environment.