AAAI 2024 论文列表

Graph clustering is a fundamental and challenging task in the field of graph mining where the objective is to group the nodes into clusters taking into consideration the topology of the graph. It has several applications in diverse domains spanning social network analysis, recommender systems, computer vision, and bioinformatics. In this work, we propose a novel method, DGCluster, which primarily optimizes the modularity objective using graph neural networks and scales linearly with the graph size. Our method does not require the number of clusters to be specified as a part of the input and can also leverage the availability of auxiliary node level information. We extensively test DGCluster on several real-world datasets of varying sizes, across multiple popular cluster quality metrics. Our approach consistently outperforms the state-of-the-art methods, demonstrating significant performance gains in almost all settings.

Fraud detection has increasingly become a prominent research field due to the dramatically increased incidents of fraud. The complex connections involving thousands, or even millions of nodes, present challenges for fraud detection tasks. Many researchers have developed various graph-based methods to detect fraud from these intricate graphs. However, those methods neglect two distinct characteristics of the fraud graph: the non-additivity of certain attributes and the distinguishability of grouped messages from neighbor nodes. This paper introduces the Dynamic Grouping Aggregation Graph Neural Network (DGA-GNN) for fraud detection, which addresses these two characteristics by dynamically grouping attribute value ranges and neighbor nodes. In DGA-GNN, we initially propose the decision tree binning encoding to transform non-additive node attributes into bin vectors. This approach aligns well with the GNN’s aggregation operation and avoids nonsensical feature generation. Furthermore, we devise a feedback dynamic grouping strategy to classify graph nodes into two distinct groups and then employ a hierarchical aggregation. This method extracts more discriminative features for fraud detection tasks. Extensive experiments on five datasets suggest that our proposed method achieves a 3% ~ 16% improvement over existing SOTA methods. Code is available at https://github.com/AtwoodDuan/DGA-GNN.

With the help of Vision transformers (ViTs), medical image segmentation was able to achieve outstanding performance. In particular, they overcome the limitation of convolutional neural networks (CNNs) which rely on local receptive fields. ViTs use self-attention mechanisms to consider relationships between all image pixels or patches simultaneously. However, they require large datasets for training and did not perform well on capturing low-level features. To that end, we propose DDViT, a novel ViT model that unites a CNN to alleviate data-hunger for medical image segmentation with two multi-scale feature representations. Significantly, our approach incorporates a ViT with a plug-in domain adapter (DA) with Double-Level Fusion (DLF) technique, complemented by a mutual knowledge distillation paradigm, facilitating the seamless exchange of knowledge between a universal network and specialized domain-specific network branches. The DLF framework plays a pivotal role in our encoder-decoder architecture, combining the innovation of the TransFuse module with a robust CNN-based encoder. Extensive experimentation across diverse medical image segmentation datasets underscores the remarkable efficacy of DDViT when compared to alternative approaches based on CNNs and Transformer-based models.

Diffusion-based generative models have recently exhibited powerful generative performance. However, as many attributes exist in the data distribution and owing to several limitations of sharing the model parameters across all levels of the generation process, it remains challenging to control specific styles for each attribute. To address the above problem, we introduce decoupled denoising diffusion models (DDDMs) with disentangled representations, which can enable effective style transfers for each attribute in generative models. In particular, we apply DDDMs for voice conversion (VC) tasks, tackling the intricate challenge of disentangling and individually transferring each speech attributes such as linguistic information, intonation, and timbre. First, we use a self-supervised representation to disentangle the speech representation. Subsequently, the DDDMs are applied to resynthesize the speech from the disentangled representations for style transfer with respect to each attribute. Moreover, we also propose the prior mixup for robust voice style transfer, which uses the converted representation of the mixed style as a prior distribution for the diffusion models. The experimental results reveal that our method outperforms publicly available VC models. Furthermore, we show that our method provides robust generative performance even when using a smaller model size. Audio samples are available at https://hayeong0.github.io/DDDM-VC-demo/.

Recently, masked image modeling (MIM) has demonstrated promising prospects in self-supervised representation learning. However, existing MIM frameworks recover all masked patches equivalently, ignoring that the reconstruction difficulty of different patches can vary sharply due to their diverse distance from visible patches. In this paper, we propose a novel deep dynamic supervision to enable MIM methods to dynamically reconstruct patches with different degrees of difficulty at different pretraining phases and depths of the model. Our deep dynamic supervision helps to provide more locality inductive bias for ViTs especially in deep layers, which inherently makes up for the absence of local prior for self-attention mechanism. Built upon the deep dynamic supervision, we propose Deep Dynamic AutoEncoder (DDAE), a simple yet effective MIM framework that utilizes dynamic mechanisms for pixel regression and feature self-distillation simultaneously. Extensive experiments across a variety of vision tasks including ImageNet classification, semantic segmentation on ADE20K and object detection on COCO demonstrate the effectiveness of our approach.

Visual interpretation is extremely important in human geography as the primary technique for geographers to use photograph data in identifying, classifying, and quantifying geographic and topological objects or regions. However, it is also time-consuming and requires overwhelming manual effort from professional geographers. This paper describes our interdisciplinary team's efforts in integrating computer vision models with geographers' visual image interpretation process to reduce their workload in interpreting images. Focusing on the dune segmentation task, we proposed an approach featuring a deep dune segmentation model to identify dunes and label their ranges in an automated way. By developing a tool to connect our model with ArcGIS, one of the most popular workbenches for visual interpretation, geographers can further refine the automatically-generated dune segmentation on images without learning any CV or deep learning techniques. Our approach thus realized a non-invasive change to geographers' visual interpretation routines, reducing their manual efforts while incurring minimal interruptions to their work routines and tools they are familiar with. Deployment with a leading Chinese geography research institution demonstrated the potential of our approach in supporting geographers in researching and solving drylands desertification.

Neural predictors have shown great potential in the evaluation process of neural architecture search (NAS). However, current predictor-based approaches overlook the fact that training a predictor necessitates a considerable number of trained neural networks as the labeled training set, which is costly to obtain. Therefore, the critical issue in utilizing predictors for NAS is to train a high-performance predictor using as few trained neural networks as possible. Although some methods attempt to address this problem through unsupervised learning, they often result in inaccurate predictions. We argue that the unsupervised tasks intended for the common graph data are too challenging for neural networks, causing unsupervised training to be susceptible to performance crashes in NAS. To address this issue, we propose a CurricuLum-guided Contrastive Learning framework for neural Predictor (DCLP). Our method simplifies the contrastive task by designing a novel curriculum to enhance the stability of unlabeled training data distribution during contrastive training. Specifically, we propose a scheduler that ranks the training data according to the contrastive difficulty of each data and then inputs them to the contrastive learner in order. This approach concentrates the training data distribution and makes contrastive training more efficient. By using our method, the contrastive learner incrementally learns feature representations via unsupervised data on a smooth learning curve, avoiding performance crashes that may occur with excessively variable training data distributions. We experimentally demonstrate that DCLP has high accuracy and efficiency compared with existing predictors, and shows promising potential to discover superior architectures in various search spaces when combined with search strategies. Our code is available at: https://github.com/Zhengsh123/DCLP.

Neural architecture search-based multi-modal classification (NAS-MMC) methods can individually obtain the optimal classifier for different multi-modal data sets in an automatic manner. However, most existing NAS-MMC methods are dramatically time consuming due to the requirement for training and evaluating enormous models. In this paper, we propose an efficient evolutionary-based NAS-MMC method called divide-and-conquer neural architecture search (DC-NAS). Specifically, the evolved population is first divided into k+1 sub-populations, and then k sub-populations of them evolve on k small-scale data sets respectively that are obtained by splitting the entire data set using the k-fold stratified sampling technique; the remaining one evolves on the entire data set. To solve the sub-optimal fusion model problem caused by the training strategy of partial data, two kinds of sub-populations that are trained using partial data and entire data exchange the learned knowledge via two special knowledge bases. With the two techniques mentioned above, DC-NAS achieves the training time reduction and classification performance improvement. Experimental results show that DC-NAS achieves the state-of-the-art results in term of classification performance, training efficiency and the number of model parameters than the compared NAS-MMC methods on three popular multi-modal tasks including multi-label movie genre classification, action recognition with RGB and body joints and dynamic hand gesture recognition.

As an up-and-coming area, CLIP-based pre-trained vision-language models can readily facilitate downstream tasks through the zero-shot or few-shot fine-tuning manners. However, they still face critical challenges in test-time generalization due to the shifts between the training and test data distributions, hindering the further improvement of the performance. To address this crucial problem, the latest works have introduced Test-Time Adaptation (TTA) techniques to CLIP which dynamically learn text prompts using only test samples. However, their limited learning capacity due to the overlook of visual modality information, and the underutilization of knowledge in previously seen test samples result in reduced performance. In this paper, we propose a novel Dual-modal Adaptive online prompting and knowledge ReTention method called DART to overcome these challenges. To increase the learning capacity, DART captures knowledge from each test sample by learning class-specific text prompts and instance-level image prompts. Additionally, to fully leverage the knowledge from previously seen test samples, DART utilizes dual-modal knowledge retention prompts to adaptively retain the acquired knowledge, thereby enhancing the predictions on subsequent test samples. Extensive experiments on various large-scale benchmarks demonstrate the effectiveness of our proposed DART against state-of-the-art methods.

3D object detection achieves good detection performance in autonomous driving. However, it requires substantial computational resources, which prevents its practical application. 2D object detection has less computational burden but lacks spatial and geometric information embedded in depth. Therefore, we present DALDet, an efficient depth-aware learning based 2D detector, achieving high-performance object detection for autonomous driving. We design an efficient one-stage detection framework and seamlessly integrate depth cues into convolutional neural network by introducing depth-aware convolution and depth-aware average pooling, which effectively improve the detector's ability to perceive 3D space. Moreover, we propose a depth-guided loss function for training DALDet, which effectively improves the localization ability of the detector. Due to the use of depth map, DALDet can also output the distance of the object, which is of great importance for driving applications such as obstacle avoidance. Extensive experiments demonstrate the superiority and efficiency of DALDet. In particular, our DALDet ranks 1st on both KITTI Car and Cyclist 2D detection test leaderboards among all 2D detectors with high efficiency as well as yielding competitive performance among many leading 3D detectors. Code will be available at https://github.com/hukefy/DALDet.

Propagation models in social networks are critical, with extensive applications across various fields and downstream tasks. However, existing propagation models are often oversimplified, scenario-specific, and lack real-world user social attributes. These limitations detaching from real-world analysis lead to inaccurate representations of the propagation process in social networks. To address these issues, we propose a User Features Attention-based DAG-Aware Variational Autoencoder (DAVA) for propagation graph generation. First, nearly 1 million pieces of user attributes data are collected. Then DAVA can integrate the analysis of propagation graph topology and corresponding user attributes as prior knowledge. By leveraging a lightweight attention-based framework and a sliding window mechanism based on BFS permutations weighted by user influence, DAVA significantly enhances the ability to generate realistic, large-scale propagation data, yielding graph scales ten times greater than those produced by existing SOTA methods. Every module of DAVA has flexibility and extension that allows for easy substitution to suit other generation tasks. Additionally, we provide a comprehensive evaluation of DAVA, one focus is the effectiveness of generated data in improving the performance of downstream tasks. During the generation process, we discover the Credibility Erosion Effect by modifying the generation rules, revealing a social phenomenon in social network propagation.

Multi-Source cross-lingual transfer learning deals with the transfer of task knowledge from multiple labelled source languages to an unlabeled target language under the language shift. Existing methods typically focus on weighting the predictions produced by language-specific classifiers of different sources that follow a shared encoder. However, all source languages share the same encoder, which is updated by all these languages. The extracted representations inevitably contain different source languages' information, which may disturb the learning of the language-specific classifiers. Additionally, due to the language gap, language-specific classifiers trained with source labels are unable to make accurate predictions for the target language. Both facts impair the model's performance. To address these challenges, we propose a Disentangled and Adaptive Network ~(DA-Net). Firstly, we devise a feedback-guided collaborative disentanglement method that seeks to purify input representations of classifiers, thereby mitigating mutual interference from multiple sources. Secondly, we propose a class-aware parallel adaptation method that aligns class-level distributions for each source-target language pair, thereby alleviating the language pairs' language gap. Experimental results on three different tasks involving 38 languages validate the effectiveness of our approach.

To enhance the efficacy of multi-scenario services in industrial recommendation systems, the emergence of multi-domain recommendation has become prominent, which entails simultaneous modeling of all domains through a unified model, effectively capturing commonalities and differences among them. However, current methods rely on manual domain partitioning, which overlook the intricate domain relationships and the heterogeneity of different domains during joint optimization, hindering the integration of domain commonalities and differences. To address these challenges, this paper proposes a universal and flexible framework D3 aimed at optimizing the multi-domain recommendation pipeline from three key aspects. Firstly, an attention-based domain adaptation module is introduced to automatically identify and incorporate domain-sensitive features during training. Secondly, we propose a fusion gate module that enables the seamless integration of commonalities and diversities among domains, allowing for implicit characterization of intricate domain relationships. Lastly, we tackle the issue of joint optimization by deriving loss weights from two complementary viewpoints: domain complexity and domain specificity, alleviating inconsistencies among different domains during the training phase. Experiments on three public datasets demonstrate the effectiveness and superiority of our proposed framework. In addition, D3 has been implemented on a real-life, high-traffic internet platform catering to millions of users daily.

Image-based virtual try-on aims to transfer a target clothing onto a specific person. A significant challenge is arbitrarily matched clothing and person lack corresponding ground truth to supervised learning. A recent pioneering work leveraged an improved cycleGAN to enable one network to generate the desired image for another network during training. However, there is no difference in the result distribution before and after the clothing changes. Therefore, using two different networks is unnecessary and may even increase the difficulty of convergence. Furthermore, the introduced human parsing used to provide body structure information in the input also have a negative impact on the try-on result. How to employ a single network for supervised learning while eliminating human parsing? To tackle these issues, we present a Cycle mapping Virtual Try-On Network (CycleVTON), which can produce photo-realistic try-on results by using a cycle mapping framework without the parser. In particular, we introduce a flow constraint loss to achieve supervised learning of arbitrarily matched clothing and person as inputs to the deformer, thus naturally mimicking the interaction between clothing and the human body. Additionally, we design a skin generation strategy that can adapt to the shape of the target clothing by dynamically adjusting the skin region, i.e., by first removing and then filling skin areas. Extensive experiments conducted on challenging benchmarks demonstrate that our proposed method exhibits superior performance compared to state-of-the-art methods.

We present that visual grounding and image captioning, which perform as two mutually inverse processes, can be bridged together for collaborative training by careful designs. By consolidating this idea, we introduce CyCo, a cyclic-consistent learning framework to ameliorate the independent training pipelines of visual grounding and image captioning. The proposed framework (1) allows the semi-weakly supervised training of visual grounding; (2) improves the performance of fully supervised visual grounding; (3) yields a general captioning model that can describe arbitrary image regions. Extensive experiments show that our fully supervised grounding model achieves state-of-the-art performance, and the semi-weakly supervised one also exhibits competitive performance compared to the fully supervised counterparts. Our image captioning model has the capability to freely describe image regions and meanwhile shows impressive performance on prevalent captioning benchmarks.

Multi-source domain adaptation (MSDA) aims to transfer knowledge from multiple source domains to the unlabeled target domain. In this paper, we propose a cycle self-refinement domain adaptation method, which progressively attempts to learn the dominant transferable knowledge in each source domain in a cycle manner. Specifically, several source-specific networks and a domain-ensemble network are adopted in the proposed method. The source-specific networks are adopted to provide the dominant transferable knowledge in each source domain for instance-level ensemble on predictions of the samples in target domain. Then these samples with high-confidence ensemble predictions are adopted to refine the domain-ensemble network. Meanwhile, to guide each source-specific network to learn more dominant transferable knowledge, we force the features of the target domain from the domain-ensemble network and the features of each source domain from the corresponding source-specific network to be aligned with their predictions from the corresponding networks. Thus the adaptation ability of source-specific networks and the domain-ensemble network can be improved progressively. Extensive experiments on Office-31, Office-Home and DomainNet show that the proposed method outperforms the state-of-the-art methods for most tasks.

Building a skilled cybersecurity workforce is paramount to building a safer digital world. However, the diverse skill set, constantly emerging vulnerabilities, and deployment of new cyber threats make learning cybersecurity challenging. Traditional education methods struggle to cope with cybersecurity's rapidly evolving landscape and keep students engaged and motivated. Different studies on students' behaviors show that an interactive mode of education by engaging through a question-answering system or dialoguing is one of the most effective learning methodologies. There is a strong need to create advanced AI-enabled education tools to promote interactive learning in cybersecurity. Unfortunately, there are no publicly available standard question-answer datasets to build such systems for students and novice learners to learn cybersecurity concepts, tools, and techniques. The education course material and online question banks are unstructured and need to be validated and updated by domain experts, which is tedious when done manually. In this paper, we propose CyberGen, a novel unification of large language models (LLMs) and knowledge graphs (KG) to generate the questions and answers for cybersecurity automatically. Augmenting the structured knowledge from knowledge graphs in prompts improves factual reasoning and reduces hallucinations in LLMs. We used the knowledge triples from cybersecurity knowledge graphs (AISecKG) to design prompts for ChatGPT and generate questions and answers using different prompting techniques. Our question-answer dataset, CyberQ, contains around 4k pairs of questions and answers. The domain expert manually evaluated the random samples for consistency and correctness. We train the generative model using the CyberQ dataset for question answering task.

Low-level vision plays a crucial role in a wide range of imaging quality and image recognition applications. However, the limited size, quality, and diversity of datasets often pose significant challenges for low-level tasks. Data augmentation is the most effective and practical way of sample expansion, but the commonly used augmentation methods in high-level tasks have limited improvement in the low-level due to the boundary effects or the non-realistic context information. In this paper, we propose the Cut-and-Swap Frequency Components (CutFreq) method for low-level vision, which aims to preserve high-level representations with directionality and improve image synthesis quality. Observing the significant frequency domain differences between reconstructed images and real ones, in CutFreq, we propose to transform the input and real images separately in the frequency domain, then define two stages for the model training process, and finally swap the specified frequency bands respectively and inversely transform to generate augmented samples. The experimental results show the superior performance of CutFreq on five low-level vision tasks. Moreover, we demonstrate the effectiveness of CutFreq in the low-data regime. Code is available at https://github.com/DreamerCCC/CutFreq.

Large language models (LLMs) are becoming increasingly important for machine learning applications. However, it can be challenging to align LLMs with our intent, particularly when we want to generate content that is preferable over others or when we want the LLM to respond in a certain style or tone that is hard to describe. To address this challenge, we propose an approach that uses contrastive examples to better describe our intent. This involves providing positive examples that illustrate the true intent, along with negative examples that show what characteristics we want LLMs to avoid. The negative examples can be retrieved from labeled data, written by a human, or generated by the LLM itself. Before generating an answer, we ask the model to analyze the examples to teach itself what to avoid. This reasoning step provides the model with the appropriate articulation of the user's need and guides it towards generting a better answer. We tested our approach on both synthesized and real-world datasets, including StackExchange and Reddit, and found that it significantly improves performance compared to standard few-shot prompting.