Xiaofeng Zhang

Zeyu Huang, Yikang Shen, Xiaofeng Zhang et al. · tencent-ai

Large Transformer-based Pretrained Language Models (PLMs) dominate almost all Natural Language Processing (NLP) tasks. Nevertheless, they still make mistakes from time to time. For a model deployed in an industrial environment, fixing these mistakes quickly and robustly is vital to improve user experiences. Previous works formalize such problems as Model Editing (ME) and mostly focus on fixing one mistake. However, the one-mistake-fixing scenario is not an accurate abstraction of the real-world challenge. In the deployment of AI services, there are ever-emerging mistakes, and the same mistake may recur if not corrected in time. Thus a preferable solution is to rectify the mistakes as soon as they appear nonstop. Therefore, we extend the existing ME into Sequential Model Editing (SME) to help develop more practical editing methods. Our study shows that most current ME methods could yield unsatisfying results in this scenario. We then introduce Transformer-Patcher, a novel model editor that can shift the behavior of transformer-based models by simply adding and training a few neurons in the last Feed-Forward Network layer. Experimental results on both classification and generation tasks show that Transformer-Patcher can successively correct up to thousands of errors (Reliability) and generalize to their equivalent inputs (Generality) while retaining the model's accuracy on irrelevant inputs (Locality). Our method outperforms previous fine-tuning and HyperNetwork-based methods and achieves state-of-the-art performance for Sequential Model Editing (SME). The code is available at https://github.com/ZeroYuHuang/Transformer-Patcher.

Xiaofeng Zhang, Yikang Shen, Zeyu Huang et al. · tencent-ai

Mixture-of-Experts (MoE) networks have been proposed as an efficient way to scale up model capacity and implement conditional computing. However, the study of MoE components mostly focused on the feedforward layer in Transformer architecture. This paper proposes the Mixture of Attention Heads (MoA), a new architecture that combines multi-head attention with the MoE mechanism. MoA includes a set of attention heads that each has its own set of parameters. Given an input, a router dynamically selects a subset of $k$ attention heads per token. This conditional computation schema allows MoA to achieve stronger performance than the standard multi-head attention layer. Furthermore, the sparsely gated MoA can easily scale up the number of attention heads and the number of parameters while preserving computational efficiency. In addition to the performance improvements, MoA also automatically differentiates heads' utilities, providing a new perspective to discuss the model's interpretability. We conducted experiments on several important tasks, including Machine Translation and Masked Language Modeling. Experiments have shown promising results on several tasks against strong baselines that involve large and very deep models.

Linhao Luo, Yixiang Fang, Moli Lu et al.

Heterogeneous graphs, which contain nodes and edges of multiple types, are prevalent in various domains, including bibliographic networks, social media, and knowledge graphs. As a fundamental task in analyzing heterogeneous graphs, relevance measure aims to calculate the relevance between two objects of different types, which has been used in many applications such as web search, recommendation, and community detection. Most of existing relevance measures focus on homogeneous networks where objects are of the same type, and a few measures are developed for heterogeneous graphs, but they often need the pre-defined meta-path. Defining meaningful meta-paths requires much domain knowledge, which largely limits their applications, especially on schema-rich heterogeneous graphs like knowledge graphs. Recently, the Graph Neural Network (GNN) has been widely applied in many graph mining tasks, but it has not been applied for measuring relevance yet. To address the aforementioned problems, we propose a novel GNN-based relevance measure, namely GSim. Specifically, we first theoretically analyze and show that GNN is effective for measuring the relevance of nodes in the graph. We then propose a context path-based graph neural network (CP-GNN) to automatically leverage the semantics in heterogeneous graphs. Moreover, we exploit CP-GNN to support relevance measures between two objects of any type. Extensive experiments demonstrate that GSim outperforms existing measures.

Ziao Wang, Yuhang Li, Junda Wu et al.

In this paper, we propose FinVis-GPT, a novel multimodal large language model (LLM) specifically designed for financial chart analysis. By leveraging the power of LLMs and incorporating instruction tuning and multimodal capabilities, FinVis-GPT is capable of interpreting financial charts and providing valuable analysis. To train FinVis-GPT, a financial task oriented dataset was generated for pre-training alignment and instruction tuning, comprising various types of financial charts and their corresponding descriptions. We evaluate the model performance via several case studies due to the time limit, and the promising results demonstrated that FinVis-GPT is superior in various financial chart related tasks, including generating descriptions, answering questions and predicting future market trends, surpassing existing state-of-the-art multimodal LLMs. The proposed FinVis-GPT serves as a pioneering effort in utilizing multimodal LLMs in the finance domain and our generated dataset will be release for public use in the near future to speedup related research.

Ziao Wang, Jianning Wang, Junda Wu et al.

At the beginning era of large language model, it is quite critical to generate a high-quality financial dataset to fine-tune a large language model for financial related tasks. Thus, this paper presents a carefully designed data creation pipeline for this purpose. Particularly, we initiate a dialogue between an AI investor and financial expert using ChatGPT and incorporate the feedback of human financial experts, leading to the refinement of the dataset. This pipeline yielded a robust instruction tuning dataset comprised of 103k multi-turn chats. Extensive experiments have been conducted on this dataset to evaluate the model's performance by adopting an external GPT-4 as the judge. The promising experimental results verify that our approach led to significant advancements in generating accurate, relevant, and financial-style responses from AI models, and thus providing a powerful tool for applications within the financial sector.

Xiaofeng Zhang, Zishan Xu, Hao Tang et al.

Low-light image enhancement is a crucial visual task, and many unsupervised methods tend to overlook the degradation of visible information in low-light scenes, which adversely affects the fusion of complementary information and hinders the generation of satisfactory results. To address this, our study introduces "Enlighten-Your-Voice", a multimodal enhancement framework that innovatively enriches user interaction through voice and textual commands. This approach does not merely signify a technical leap but also represents a paradigm shift in user engagement. Our model is equipped with a Dual Collaborative Attention Module (DCAM) that meticulously caters to distinct content and color discrepancies, thereby facilitating nuanced enhancements. Complementarily, we introduce a Semantic Feature Fusion (SFM) plug-and-play module that synergizes semantic context with low-light enhancement operations, sharpening the algorithm's efficacy. Crucially, "Enlighten-Your-Voice" showcases remarkable generalization in unsupervised zero-shot scenarios. The source code can be accessed from https://github.com/zhangbaijin/Enlighten-Your-Voice

Kai Zhang, Xingyu Chen, Xiaofeng Zhang

Large Multimodal Models (LMMs) have become a pivotal research focus in deep learning, demonstrating remarkable capabilities in 3D scene understanding. However, current 3D LMMs employing thousands of spatial tokens for multimodal reasoning suffer from critical inefficiencies: excessive computational overhead and redundant information flows. Unlike 2D VLMs processing single images, 3D LMMs exhibit inherent architectural redundancy due to the heterogeneous mechanisms between spatial tokens and visual tokens. To address this challenge, we propose AdaToken-3D, an adaptive spatial token optimization framework that dynamically prunes redundant tokens through spatial contribution analysis. Our method automatically tailors pruning strategies to different 3D LMM architectures by quantifying token-level information flows via attention pattern mining. Extensive experiments on LLaVA-3D (a 7B parameter 3D-LMM) demonstrate that AdaToken-3D achieves 21\% faster inference speed and 63\% FLOPs reduction while maintaining original task accuracy. Beyond efficiency gains, this work systematically investigates redundancy patterns in multimodal spatial information flows through quantitative token interaction analysis. Our findings reveal that over 60\% of spatial tokens contribute minimally ($<$5\%) to the final predictions, establishing theoretical foundations for efficient 3D multimodal learning.

Yizhang Wang, Di Wang, You Zhou et al.

The widely applied density peak clustering (DPC) algorithm makes an intuitive cluster formation assumption that cluster centers are often surrounded by data points with lower local density and far away from other data points with higher local density. However, this assumption suffers from one limitation that it is often problematic when identifying clusters with lower density because they might be easily merged into other clusters with higher density. As a result, DPC may not be able to identify clusters with variational density. To address this issue, we propose a variational density peak clustering (VDPC) algorithm, which is designed to systematically and autonomously perform the clustering task on datasets with various types of density distributions. Specifically, we first propose a novel method to identify the representatives among all data points and construct initial clusters based on the identified representatives for further analysis of the clusters' property. Furthermore, we divide all data points into different levels according to their local density and propose a unified clustering framework by combining the advantages of both DPC and DBSCAN. Thus, all the identified initial clusters spreading across different density levels are systematically processed to form the final clusters. To evaluate the effectiveness of the proposed VDPC algorithm, we conduct extensive experiments using 20 datasets including eight synthetic, six real-world and six image datasets. The experimental results show that VDPC outperforms two classical algorithms (i.e., DPC and DBSCAN) and four state-of-the-art extended DPC algorithms.

Liqi Yang, Linhan Luo, Lifeng Xin et al.

Session-based recommendations have been widely adopted for various online video and E-commerce Websites. Most existing approaches are intuitively proposed to discover underlying interests or preferences out of the anonymous session data. This apparently ignores the fact these sequential behaviors usually reflect session user's potential demand, i.e., a semantic level factor, and therefore how to estimate underlying demands from a session is challenging. To address aforementioned issue, this paper proposes a demand-aware graph neural networks (DAGNN). Particularly, a demand modeling component is designed to first extract session demand and the underlying multiple demands of each session is estimated using the global demand matrix. Then, the demand-aware graph neural network is designed to extract session demand graph to learn the demand-aware item embedddings for the later recommendations. The mutual information loss is further designed to enhance the quality of the learnt embeddings. Extensive experiments are evaluated on several real-world datasets and the proposed model achieves the SOTA model performance.

Shiyi Chen, Ziao Wang, Xinni Zhang et al.

Graph representation learning has long been an important yet challenging task for various real-world applications. However, their downstream tasks are mainly performed in the settings of supervised or semi-supervised learning. Inspired by recent advances in unsupervised contrastive learning, this paper is thus motivated to investigate how the node-wise contrastive learning could be performed. Particularly, we respectively resolve the class collision issue and the imbalanced negative data distribution issue. Extensive experiments are performed on three real-world datasets and the proposed approach achieves the SOTA model performance.

Wenxin Hu, Xiaofeng Zhang, Gang Yang

Automatically generating macro research reports from economic news is an important yet challenging task. As we all know, it requires the macro analysts to write such reports within a short period of time after the important economic news are released. This motivates our work, i.e., using AI techniques to save manual cost. The goal of the proposed system is to generate macro research reports as the draft for macro analysts. Essentially, the core challenge is the long text generation issue. To address this issue, we propose a novel deep learning technique based approach which includes two components, i.e., outline generation and macro research report generation.For the model performance evaluation, we first crawl a large news-to-report dataset and then evaluate our approach on this dataset, and the generated reports are given for the subjective evaluation.

Xin He, Shihao Wang, Shaohuai Shi et al.

Skin disease is one of the most common types of human diseases, which may happen to everyone regardless of age, gender or race. Due to the high visual diversity, human diagnosis highly relies on personal experience; and there is a serious shortage of experienced dermatologists in many countries. To alleviate this problem, computer-aided diagnosis with state-of-the-art (SOTA) machine learning techniques would be a promising solution. In this paper, we aim at understanding the performance of convolutional neural network (CNN) based approaches. We first build two versions of skin disease datasets from Internet images: (a) Skin-10, which contains 10 common classes of skin disease with a total of 10,218 images; (b) Skin-100, which is a larger dataset that consists of 19,807 images of 100 skin disease classes. Based on these datasets, we benchmark several SOTA CNN models and show that the accuracy of skin-100 is much lower than the accuracy of skin-10. We then implement an ensemble method based on several CNN models and achieve the best accuracy of 79.01\% for Skin-10 and 53.54\% for Skin-100. We also present an object detection based approach by introducing bounding boxes into the Skin-10 dataset. Our results show that object detection can help improve the accuracy of some skin disease classes.

Dan Peng, Zizhan Zheng, Xiaofeng Zhang

Adversarial examples are perturbed inputs designed to fool machine learning models. Most recent works on adversarial examples for image classification focus on directly modifying pixels with minor perturbations. A common requirement in all these works is that the malicious perturbations should be small enough (measured by an L_p norm for some p) so that they are imperceptible to humans. However, small perturbations can be unnecessarily restrictive and limit the diversity of adversarial examples generated. Further, an L_p norm based distance metric ignores important structure patterns hidden in images that are important to human perception. Consequently, even the minor perturbation introduced in recent works often makes the adversarial examples less natural to humans. More importantly, they often do not transfer well and are therefore less effective when attacking black-box models especially for those protected by a defense mechanism. In this paper, we propose a structure-preserving transformation (SPT) for generating natural and diverse adversarial examples with extremely high transferability. The key idea of our approach is to allow perceptible deviation in adversarial examples while keeping structure patterns that are central to a human classifier. Empirical results on the MNIST and the fashion-MNIST datasets show that adversarial examples generated by our approach can easily bypass strong adversarial training. Further, they transfer well to other target models with no loss or little loss of successful attack rate.

Zhaocai Sun, William K. Cheung, Xiaofeng Zhang et al.

Semi-supervised learning plays an important role in large-scale machine learning. Properly using additional unlabeled data (largely available nowadays) often can improve the machine learning accuracy. However, if the machine learning model is misspecified for the underlying true data distribution, the model performance could be seriously jeopardized. This issue is known as model misspecification. To address this issue, we focus on generative models and propose a criterion to detect the onset of model misspecification by measuring the performance difference between models obtained using supervised and semi-supervised learning. Then, we propose to automatically modify the generative models during model training to achieve an unbiased generative model. Rigorous experiments were carried out to evaluate the proposed method using two image classification data sets PASCAL VOC'07 and MIR Flickr. Our proposed method has been demonstrated to outperform a number of state-of-the-art semi-supervised learning approaches for the classification task.