Bo Pan, Zhen Xiong, Guanchen Wu, Zheng Zhang, Yifei Zhang, Yuntong Hu, Liang Zhao. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Text-Attributed Graphs (TAGs) enhance graph structures with natural language descriptions, enabling detailed representation of data and their relationships across a broad spectrum of real-world scenarios. Despite the potential for deeper insights, existing TAG representation learning primarily omit the semantic relationship among node texts, and mostly relies on supervised methods, necessitating extensive labeled data and limiting applicability across diverse contexts. This paper introduces a new self-supervised learning framework, Text-Attributed-Graph Multi-View Alignment (TAGA), which overcomes these constraints by integrating TAGs' structural and semantic dimensions. TAGA constructs two complementary views: Text-of-Graph view, which organizes node texts into structured documents based on graph topology, and the Graph-of-Text view, which converts textual nodes and connections into graph data. By aligning representations from both views, TAGA captures joint textual and structural information. In addition, a novel structure-preserving random walk algorithm is proposed for efficient training on large-sized TAGs. Our framework demonstrates strong performance in zero-shot and few-shot scenarios across eight real-world datasets.

Explainable AI (XAI) has gained significant attention for providing insights into the decision-making processes of deep learning models, particularly for image classification tasks through saliency-based visual explanations. Despite their success, key challenges persist due to the scarcity of annotated datasets and the absence of standardized evaluation protocols. In this paper, we introduce Saliency-Bench, a novel benchmark designed to evaluate visual explanations generated by saliency methods across multiple datasets. We curated, constructed, and annotated eight datasets, each covering diverse tasks such as scene classification, cancer diagnosis, object classification, and action classification, with corresponding ground-truth explanation annotations. The benchmark includes a standardized and unified evaluation pipeline for assessing faithfulness and alignment of the visual explanation, providing a holistic visual explanation performance assessment. We benchmark these eight datasets with widely used saliency methods on different image classifier architectures to evaluate explanation quality. Additionally, we developed an user-friendly toolkit for automating the evaluation pipeline, from data accessing, and data loading, to result evaluation. The benchmark is available at https://github.com/XAIdataset/XAIdataset.github.io.

This paper presents a novel Chemist-AI Alignment framework for generating novel structures of quaternary ammonium compounds (QACs), a crucial class of antimicrobial agents.The framework uniquely integrates AI-driven small molecule generation with iterative feedback from chemist experts, leveraging both rapid assessments and comprehensive wet-lab validations to optimize for biological potency and synthetic feasibility.Central to the framework is a hierarchical generative model that captures the QAC hierarchical topology.Extensive experiments highlight the efficacy of the framework in identifying promising QAC candidates, many

Text-Attributed Graphs (TAGs) are graphs of connected textual documents. Graph models can efficiently learn TAGs, but their training heavily relies on human-annotated labels, which are scarce or even unavailable in many applications. Large language models (LLMs) have recently demonstrated remarkable capabilities in few-shot and zero-shot TAG learning, but they suffer from scalability, cost, and privacy issues. Therefore, in this work, we focus on synergizing LLMs and graph models with their complementary strengths by distilling the power of LLMs into a local graph model on TAG learning. To address the inherent gaps between LLMs (generative models for texts) and graph models (discriminative models for graphs), we propose first to let LLMs teach an interpreter with rich rationale and then let a student model mimic the interpreter's reasoning without LLMs' rationale. We convert LLM's textual rationales to multi-level graph rationales to train the interpreter model and align the student model with the interpreter model based on the features of TAGs. Extensive experiments validate the efficacy of our proposed framework.

Explanation supervision is a technique in which the model is guided by human-generated explanations during training. This technique aims to improve the predictability of the model by incorporating human understanding of the prediction process into the training phase. This is a challenging task since it relies on the accuracy of human annotation labels. To obtain high-quality explanation annotations, using multiple annotations to do explanation supervision is a reasonable method. However, how to use multiple annotations to improve accuracy is particularly challenging due to the following: 1) The noisiness of annotations from different annotators; 2) The lack of pre-given information about the corresponding relationship between annotations and annotators; 3) Missing annotations since some images are not labeled by all annotators. To solve these challenges, we propose a Multi-annotated explanation-guided learning (MAGI) framework to do explanation supervision with comprehensive and high-quality generated annotations. We first propose a novel generative model to generate annotations from all annotators and infer them using a newly proposed variational inference-based technique by learning the characteristics of each annotator. We also incorporate an alignment mechanism into the generative model to infer the correspondence between annotations and annotators in the training process. Extensive experiments on two datasets from the medical imaging domain demonstrate the effectiveness of our proposed framework in handling noisy annotations while obtaining superior prediction performance compared with previous SOTA.

Designing molecules with desired biological properties remains an outstanding challenge both in the wet and dry laboratories. Meeting this challenge promises great translational impacts across drug discovery, material sciences, biotechnology, and more. Recent momentum in deep learning promises to advance our computational capabilities on molecule generation. In particular, deep graph generative models which treat molecule design as a graph generation problem are allowing us to directly learn from existing databases of small molecules and generate novel, valid molecules. Currently, these models have many shortcomings, including poor controllability of desired molecular properties, especially in practical application where the training data is usually small, noisy, and incomplete. This paper focuses on equipping graph variational autoencoders with the ability to control for desired properties and its practical application in a practical application which is the generation of Quaternary Ammonium Compounds (QAC). Several controllable graph generation mechanisms are investigated for their effectiveness. A general framework is then proposed to extend these mechanisms by our newly proposed objective function to handle the challenges in practical applications where the property value annotations are usually censored and not fully available in all training samples. The experimental evaluation considers an experimentally-characterized dataset of antimicrobial small molecules with wet-lab characterized activity against antibiotic-resistant bacteria. Extensive experiments demonstrate the superiority of the proposed models and control of desired properties.

Activity characterization, optimization, and generation of small molecules are increasingly active areas of research at the intersection of molecular chemistry and machine learning. Large datasets of small molecules have allowed training deep models that have been shown capable of exploring the underlying chemical space and generating valid, novel, and unique molecules. While this is a noteworthy achievement, what impedes operationalizing these models in the wet laboratory is the ability to link the chemical and biological space of small molecules. A central challenge to this is the lack of activity data on these entities. In this paper we relate a computational pipeline that permits linking the chemical and biological space of an important class of small molecules, quaternary ammonium compounds (QACs). Our experimental collaborators have characterized the activity of many QACs against Staphylococcus aureus. We train various generative models and evaluate their ability to generate valid, novel, and unique QACs. We then leverage classification models trained over activity data to evaluate the generated QACs. The resulting pipeline identifies valid, novel, unique, membrane-active QACs. This work opens the way to further avenues of research in machine learning models capable of jointly sampling the chemical and biological space of small molecules.

Developing deep generative models has been an emerging field due to the ability to model and generate complex data for various purposes, such as image synthesis and molecular design. However, the advance of deep generative models is limited by the challenges to generate objects that possess multiple desired properties because: 1) the existence of complex correlation among real-world properties is common but hard to identify; 2) controlling individual property enforces an implicit partially control of its correlated properties, which is difficult to model; 3) controlling multiple properties under variour manners simultaneously is hard and underexplored. We address these challenges by proposing a novel deep generative framework that recovers semantics and correlation of properties through disentangled latent vectors. The correlation is handled via an explainable mask pooling layer, and properties are precisely retained by the generated objects via the mutual dependence between latent vectors and properties. Our generative model preserves properties of interest while handles correlation and conflicts of properties under a multi-objective optimization framework. The experiments demonstrate our model's superior performance in generating objects with desired properties.