The impressive performance of large language models (LLMs) also brings inherent toxicity risks, prompting the need for effective detoxification to support responsible deployment. Prevailing methods generally follow an inflexible model-specific fashion, addressing only individual models or model families. Moreover, overlooking the underlying toxic risks involved in the input prefix can lead to toxic accumulation during autoregressive generation. Existing methods rely on external strong attribute interventions to address this issue, which further exacerbates contextual semantic inconsistencies and makes it difficult to balance toxicity efficacy and generation quality. To address these concerns, we propose a novel Model-Agnostic Adaptive Detoxification (MAAD) framework. To address accumulating toxicity, we present prefix heuristics that serve as contextual signals, guiding the base LLM toward safer generation. Along this line, we construct an antidote dataset to support a lightweight model, Detoxifier, which steers the base LLM to make in-scope and reliable detoxifying distribution adjustments while preserving fluency and contextual understanding. Designed as an easy-to-deploy module, Detoxifier requires a small amount of data and can be seamlessly applied to various base LLMs with one-off training. Since over-purifying often reduces diversity, we also propose a dynamic truncation method called CW-cutoff sampling to trade off language model quality and diversity. Extensive experiments demonstrate that MAAD strikes a better balance between detoxification effectiveness and generation quality, while also maintaining model utility.

Brain graphs (brain connectivity networks) play an important role in modeling the complex structure of the human brain. Furthermore, brain graph learning based on graph neural networks (GNNs) has recently attracted growing interest. Although existing methods have made great progress in brain disorder prediction and pathogenic analysis, there are two key problems: (1) They rarely utilize the pathogenic reason of brain disorders, that is, salient brain regions always lead to abnormal connections between brain regions, to extract critical brain graph information for disorder analysis; (2) Since most of the available brain graph data is limited, how can we improve the performance of brain graph learning models on insufficient training data? Thus, in this paper, we learn brain graph representations for disorder prediction and analyze disorder-specific brain regions and connections from the subgraph perspective. Besides, we introduce the cross-domain brain graph learning framework to alleviate the problem of poor model performance on limited data. To consider the pathogenic reason by brain subgraphs, we first propose the node entropy of brain graphs based on brain graph properties to extract important nodes. We then introduce subgraph information bottleneck to refine the critical subgraph from the rough subgraph generated by these important nodes, recognizing important connections related to disorders. To achieve a better model performance on limited data, we design a cross-domain brain graph learning framework to improve the subgraph extraction model by the meta-learning method. The subgraph extraction model is pre-trained on a large source training dataset and then quickly adapted to target task dataset. Besides, a simple yet effective feature alignment module is applied to mitigate the negative transfer problem for cross-domain datasets. Extensive experimental results, including disorder prediction and pathogenic analysis on real-world neuroimaging data, demonstrate the effectiveness of our method.

Anomaly detection in graph-structured data is an inherently challenging problem, as it requires the identification of rare nodes that deviate from the majority in both their structural and behavioral characteristics. Existing methods, such as those based on graph convolutional networks (GCNs), often suffer from over-smoothing, which causes the learned node representations to become indistinguishable. Furthermore, graph reconstruction-based approaches are vulnerable to anomalous node interference during the reconstruction process, leading to inaccurate anomaly detection. In this work, we propose a novel and holistic anomaly evaluation framework that integrates three key components: a local-global Transformer encoder, a memory-guided reconstruction mechanism and a multi-scale representation matching strategy. These components work synergistically to enhance the model's ability to capture both local and global structural dependencies, suppress the influence of anomalous nodes, and assess anomalies from multiple levels of granularity. Anomaly scores are computed by combining reconstruction errors and memory matching signals, resulting in a more robust evaluation. Extensive experiments on seven benchmark datasets demonstrate that our method outperforms existing state-of-the-art approaches, offering a comprehensive and generalizable solution for anomaly detection across various graph domains.

The human brain is a highly complex neurological system that has been the subject of continuous exploration by scientists. With the help of modern neuroimaging techniques, there has been significant progress made in brain disorder analysis. There is an increasing interest about utilizing artificial intelligence techniques to improve the efficiency of disorder diagnosis in recent years. However, these methods rely only on neuroimaging data for disorder diagnosis and do not explore the pathogenic mechanism behind the disorder or provide an interpretable result toward the diagnosis decision. Furthermore, the scarcity of medical data limits the performance of existing methods. As the hot application of graph neural networks (GNNs) in molecular graphs and drug discovery due to its strong graph-structured data learning ability, whether GNNs can also play a huge role in the field of brain disorder analysis. Thus, in this work, we innovatively model brain neuroimaging data into graph-structured data and propose knowledge distillation (KD) guided brain subgraph neural networks to extract discriminative subgraphs between patient and healthy brain graphs to explain which brain regions and abnormal functional connectivities cause the disorder. Specifically, we introduce the KD technique to transfer the knowledge of pretrained teacher model to guide brain subgraph neural networks training and alleviate the problem of insufficient training data. And these discriminative subgraphs are conducive to learn better brain graph-level representations for disorder prediction. We conduct abundant experiments on two functional magnetic resonance imaging datasets, i.e., Parkinson's disease (PD) and attention-deficit/hyperactivity disorder (ADHD), and experimental results well demonstrate the superiority of our method over other brain graph analysis methods for disorder prediction accuracy. The interpretable experimental results given by our method are consistent with corresponding medical research, which is encouraging to provide a potential for deeper brain disorder study.

Deep learning is increasingly crucial in scientific discovery, accelerating research in various fields. Exploring brain science using deep learning has garnered significant interest, particularly in the recognition of brain disorders. However, existing methods face limitations in the discriminability of learned brain graph representations and the identification of neurological biomarkers associated with brain disorders. Moreover, the lack of model explainability leads to suboptimal brain graph analysis. To address these challenges, we propose a bi-level brain graph contrastive learning framework with an interpretable learning kernel for brain disorder analysis. Our framework diverges from traditional graph contrastive learning methods by augmenting meaningful brain graphs using the counterfactual thinking-based mechanism, ensuring reliable graph generation that preserves discriminative information. Secondly, we introduce bi-level contrastive loss with intra-and inter-class contrasts to enhance the brain graph representation learning. Most importantly, we design an interpretable brain graph learning kernel to highlight critical regions and connections, thereby facilitating the discovery of potential neurological biomarkers associated with brain disorders in subsequent analysis. The effectiveness of our method11https://githuh.com/JustinGie/Cf-BCL in brain graph representation learning and discriminative substructure detection is demonstrated through the evaluation of disorder prediction and pathogenic analysis on three real-world brain disorder datasets. Moreover, our framework may provide novel insights into brain science based on the results of comparison with medical research.

Graph-level anomaly detection (GLAD) has already gained significant importance and has become a popular field of study, attracting considerable attention across numerous downstream works. The core focus of this domain is to capture and highlight the anomalous information within given graph datasets. In most existing studies, anomalies are often the instances of few. The stark imbalance misleads current GLAD methods to focus on learning the patterns of normal graphs more, further impacting anomaly detection performance. Moreover, existing methods predominantly utilize the inherent features of nodes to identify anomalous graph patterns which is approved suboptimal according to our experiments. In this work, we propose an imbalanced GLAD method via counterfactual augmentation and feature learning. Specifically, we first construct anomalous samples based on counterfactual learning, aiming to expand and balance the datasets. Additionally, we construct a module based on Graph Neural Networks (GNNs), which allows us to utilize degree attributes to complement the inherent attribute features of nodes. Then, we design an adaptive weight learning module to integrate features tailored to different datasets effectively to avoid indiscriminately treating all features as equivalent. Furthermore, extensive baseline experiments conducted on public datasets substantiate the robustness and effectiveness. Besides, we apply the model to brain disease datasets, which can prove the generalization capability of our work. The source code of our work is available online1.

In this paper, we propose an interpretable brain graph contrastive learning framework, which aims to learn brain graph representations by a unsupervised way for disorder prediction and pathogenic analysis. Our framework consists of two key designs: We first utilize the controllable data augmentation strategy to perturb unimportant structures and attribute features for the generation of brain graphs. Then, considering that the difference of healthy and patient brain graphs is small, we introduce hard negative sample evaluation to weight negative samples of the contrastive loss, which can learn more discriminative brain graph representations. More importantly, our method can observe salient brain regions and connections for pathogenic analysis. We conduct disorder prediction and interpretable analysis experiments on three real-world neuroimaging datasets to demonstrate the effectiveness of our framework.

Graph pooling technique as the essential component of graph neural networks has gotten increasing attention recently and it aims to learn graph-level representations for the whole graph. Besides, graph pooling is important in graph classification and graph generation tasks. However, current graph pooling methods mainly coarsen a sequence of small-sized graphs to capture hierarchical structures, potentially resulting in the deterioration of the global structure of the original graph and influencing the quality of graph representations. Furthermore, these methods artificially select the number of graph pooling layers for different graph datasets rather than considering each graph individually. In reality, the structure and size differences among graphs necessitate a specific number of graph pooling layers for each graph. In this work, we propose reinforced pooling graph neural networks via adaptive hybrid graph coarsening networks. Specifically, we design a hybrid graph coarsening strategy to coarsen redundant structures of the original graph while retaining the global structure. In addition, we introduce multi-agent reinforcement learning to adaptively perform the graph coarsening process to extract the most representative coarsened graph for each graph, enhancing the quality of graph-level representations. Finally, we design graph-level contrast to improve the preservation of global information in graph-level representations. Extensive experiments with rich baselines on six benchmark datasets show the effectiveness of ReiPool1.

To reduce the repetitive and complex work of instructors, exam paper generation (EPG) technique has become a salient topic in the intelligent education field, which targets at generating high-quality exam paper automatically according to instructor-specified assessment criteria. The current advances utilize the ability of heuristic algorithms to optimize several well-known objective constraints, such as difficulty degree, number of questions, etc., for producing optimal solutions. However, in real scenarios, considering other equally relevant objectives (e.g., distribution of exam scores, skill coverage) is extremely important. Besides, how to develop an automatic multi-objective solution that finds an optimal subset of questions from a huge search space of large- sized question datasets and thus composes a high-quality exam paper is urgent but non-trivial. To this end, we skillfully design a reinforcement learning guided Multi-Objective Exam Paper Generation framework, termed MOEPG, to simultaneously optimize three exam domain-specific objectives including difficulty degree, distribution of exam scores, and skill coverage. Specifically, to accurately measure the skill proficiency of the examinee group, we first employ deep knowledge tracing to model the interaction information between examinees and response logs. We then design the flexible Exam Q-Network, a function approximator, which automatically selects the appropriate question to update the exam paper composition process. Later, MOEPG divides the decision space into multiple subspaces to better guide the updated direction of the exam paper. Through extensive experiments on two real-world datasets, we demonstrate that MOEPG is feasible in addressing the multiple dilemmas of exam paper generation scenario1. 1https://github.com/researcher-tiger/MOEPG

Graph structure patterns are widely used to model different area data recently. How to detect anomalous graph information on these graph data has become a popular research problem. The objective of this research is centered on the particular issue that how to detect abnormal graphs within a graph set. The previous works have observed that abnormal graphs mainly show node-level and graph-level anomalies, but these methods equally treat two anomaly forms above in the evaluation of abnormal graphs, which is contrary to the fact that different types of abnormal graph data have different degrees in terms of node-level and graph-level anomalies. Furthermore, abnormal graphs that have subtle differences from normal graphs are easily escaped detection by the existing methods. Thus, we propose a multi-representations space separation based graph-level anomaly-aware detection framework in this paper. To consider the different importance of node-level and graph-level anomalies, we design an anomaly-aware module to learn the specific weight between them in the abnormal graph evaluation process. In addition, we learn strictly separate normal and abnormal graph representation spaces by four types of weighted graph representations against each other including anchor normal graphs, anchor abnormal graphs, training normal graphs, and training abnormal graphs. Based on the distance error between the graph representations of the test graph and both normal and abnormal graph representation spaces, we can accurately determine whether the test graph is anomalous. Our approach has been extensively evaluated against baseline methods using ten public graph datasets, and the results demonstrate its effectiveness. The code for our method is publicly available on https://github.com/whb605/MssGAD.git

Graph anomaly detection, here, aims to find rare patterns that are significantly different from other nodes. Attributed graphs containing complex structure and attribute information are ubiquitous in our life scenarios such as bank account transaction graph and paper citation graph. Anomalous nodes on attributed graphs show great difference from others in the perspectives of structure and attributes, and give rise to various types of graph anomalies. In this paper, we investigate three types of graph anomalies: local, global, and structure anomalies. And, graph neural networks (GNNs) based anomaly detection methods attract considerable research interests due to the power of modeling attributed graphs. However, the convolution operation of GNNs aggregates neighbors information to represent nodes, which makes node representations more similar and cannot effectively distinguish between normal and anomalous nodes, thus result in sub-optimal results. To improve the performance of anomaly detection, we propose a novel community-aware attributed graph anomaly detection framework (ComGA). We design a tailored deep graph convolutional network (tGCN) to anomaly detection on attributed graphs. Extensive experiments on eight real-life graph datasets demonstrate the effectiveness of ComGA.

Network representation aims to learn low-dimensional vector representations of network nodes while preserving the inherent properties of the network. For all its popularity, majority of the existing methods focus on exploitation of diverse information, including network topology and semantic information on nodes of network, and ignore their implicit semantics. For example, we all know the saying that birds of a feather flock together. More concretely, semantic information of one node can be influenced by its neighbors' semantic information. Furthermore, even two nodes are not directly connected, they may have similar implicit semantic information (i.e., high-order semantic proximity). Thus, they should be close in the represented vector space. To this end, we propose a Deep Semantic Network Representation approach (DSNR) in the self-translation framework from sequence to sequence. To excavate the implicit semantic information of nodes and capture the high-order semantic proximity, three key components make our approach effective, i.e., aggregation of nodes neighbors' semantic information and enhancement to the semantic feature representations of nodes by a deep autoencoder, integration of nodes semantic information in node identity sequence to generate node semantic sequence, and translation from node semantic sequence to node identity sequence to capture the high-order semantic proximity in an attention-enhanced seq2seq framework. Extensive experiments based on three real-world datasets have verified the effectiveness of our proposed approach11Code is available at https://github.com/DASE4/DSNR.