Deep learning has demonstrated expert-level performance in melanoma classification, positioning it as a powerful tool in clinical dermatology. However, model opacity and the lack of interpretability remain critical barriers to clinical adoption, as clinicians often struggle to trust the decision-making processes of black-box models. To address this gap, we present a Cross-modal Explainable Framework for Melanoma (CEFM) that leverages contrastive learning as the core mechanism for achieving interpretability. Specifically, CEFM maps clinical criteria for melanoma diagnosis—namely Asymmetry, Border, and Color (ABC)—into the Vision Transformer embedding space using dual projection heads, thereby aligning clinical semantics with visual features. The aligned representations are subsequently translated into structured textual explanations via natural language generation, creating a transparent link between raw image data and clinical interpretation. Experiments on public datasets demonstrate 92.79% accuracy and an AUC of 0.961, along with significant improvements across multiple interpretability metrics. Qualitative analyses further show that the spatial arrangement of the learned embeddings aligns with clinicians’ application of the ABC rule, effectively bridging the gap between high-performance classification and clinical trust.

Yuzhen Shi, Huanghai Liu, Yiran HU, Song Gaojie, Xu Xinran, Yubo Ma, Tianyi Tang, Li Zhang, Qingjing Chen, Feng Di, Wenbo Lv, Weiheng Wu, Kexin Yang, Sen Yang, Wei Wang, Rongyao Shi, Qiu Yuanyang, Yuemeng Qi, Zhang Jingwen, Sui Xiaoyu, Yifan Chen, Zhang Yi, An Yang, Bowen Yu, Dayiheng Liu, Junyang Lin, Weixing Shen, Bing Zhao, Charles L. A. Clarke, HU Wei. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Zhaofen Wu, Hanrong Zhang, Fulin Lin, Wujiang Xu, Xinran Xu, Yankai Chen, Henry Peng Zou, Shaowen Chen, Weizhi Zhang, Xue Liu, Philip S. Yu, Hongwei Wang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Scaling the effective context length is essential for advancing large language models (LLMs) toward artificial general intelligence (AGI). However, the quadratic increase in computational complexity inherent in traditional attention mechanisms presents a prohibitive overhead. Existing approaches either impose strongly biased structures, such as sink or window attention which are task-specific, or radically modify the attention mechanism into linear approximations, whose performance in complex reasoning tasks remains inadequately explored. In this work, we propose a solution that adheres to the ``less structure'' principle, allowing the model to determine where to attend autonomously, rather than introducing predefined biases. We introduce Mixture of Block Attention (MoBA), an innovative approach that applies the principles of Mixture of Experts (MoE) to the attention mechanism. This novel architecture demonstrates superior performance on long-context tasks while offering a key advantage: the ability to seamlessly transition between full and sparse attention, enhancing efficiency without the risk of compromising performance. MoBA has already been deployed to handle actual production workloads with long-context requirements, demonstrating significant advancements in efficient attention computation for LLMs. Our code is available at https://github.com/MoonshotAI/MoBA.

Integrated sensing and communication (ISAC) is a key technology for next-generation communication systems. Accurate sensing relies on estimating the channel’s delay and Doppler shift using reference signals (RSs), enabling the determination of reflector distances and velocities. In this paper, we propose a multi-antenna RS design for orthogonal frequency division multiplexing (OFDM)-based ISAC systems with a single receiving antenna. We exploit the two-dimensional (2D) convolutional characteristics of the delay-Doppler domain channel and the quasi-orthogonality of zero correlation zone (ZCZ) sequences to efficiently separate channel responses from multiple antennas with low complexity. The proposed parameter estimation algorithm is compatible with conventional 2D-fast Fourier transform algorithms. Numerical simulations demonstrate that in single-path scenarios, the estimation performance reaches the Cramér-Rao lower bound (CRLB). However, in multipath scenarios, performance diverges from the CRLB at high signal-to-noise ratios due to interference between paths.

Abstract Direct volume rendering (DVR) is a widely used technique in the visualisation of volumetric data. As an important DVR technique, volumetric path tracing (VPT) simulates light transport to produce realistic rendering results, which provides enhanced perception and understanding for users, especially in the field of medical imaging. VPT, based on the Monte Carlo (MC) method, typically requires a large number of samples to generate noise‐free results. However, in real‐time applications, only a limited number of samples per pixel is allowed and significant noise can be created. This paper introduces a novel neural denoising approach that utilises a new feature fusion method for VPT. Our method uses a feature decomposition technique that separates radiance into components according to noise levels. Our new decomposition technique mitigates biases found in the contemporary decoupling denoising algorithm and shows better utilisation of samples. A lightweight dual‐input network is designed to correlate these components with noise‐free ground truth. Additionally, for denoising sequences of video frames, we develop a learning‐based temporal method that calculates temporal weight maps, blending reprojected results of previous frames with spatially denoised current frames. Comparative results demonstrate that our network performs faster inference than existing methods and can produce denoised output of higher quality in real time.

Drug-drug interactions (DDIs) are one of the major concerns in pharmaceutical research, and a number of computational methods have been developed to predict whether two drugs interact or not. Recently, more attention has been paid to events caused by the DDIs, which is more useful for investigating the mechanism hidden behind the combined drug usage or adverse reactions. However, some rare events may only have few examples, hindering them from being precisely predicted. To address the above issues, we present a few-shot computational method named META-DDIE, which consists of a representation module and a comparing module, to predict DDI events. We collect drug chemical structures and DDIs from DrugBank, and categorize DDI events into hundreds of types using a standard pipeline. META-DDIE uses the structures of drugs as input and learns the interpretable representations of DDIs through the representation module. Then, the model uses the comparing module to predict whether two representations are similar, and finally predicts DDI events with few labeled examples. In the computational experiments, META-DDIE outperforms several baseline methods and especially enhances the predictive capability for rare events. Moreover, META-DDIE helps to identify the key factors that may cause DDI events and reveal the relationship among different events.

MiRNAs are a class of small non-coding RNA molecules that play an important role in many biological processes, and determining miRNA-disease associations can benefit drug development and clinical diagnosis. Although great efforts have been made to develop miRNA-disease association prediction methods, few attention has been paid to in-depth classification of miRNA-disease associations, e.g. up/down-regulation of miRNAs in diseases. In this paper, we regard known miRNA-disease associations as a signed bipartite network, which has miRNA nodes, disease nodes and two types of edges representing up/down-regulation of miRNAs in diseases, and propose a signed graph neural network method (SGNNMD) for predicting deregulation types of miRNA-disease associations. SGNNMD extracts subgraphs around miRNA-disease pairs from the signed bipartite network and learns structural features of subgraphs via a labeling algorithm and a neural network, and then combines them with biological features (i.e. miRNA-miRNA functional similarity and disease-disease semantic similarity) to build the prediction model. In the computational experiments, SGNNMD achieves highly competitive performance when compared with several baselines, including the signed graph link prediction methods, multi-relation prediction methods and one existing deregulation type prediction method. Moreover, SGNNMD has good inductive capability and can generalize to miRNAs/diseases unseen during the training.

Enhancer-promoter interactions (EPIs) regulate the expression of specific genes in cells, which help facilitate understanding of gene regulation, cell differentiation and disease mechanisms. EPI identification approaches through wet experiments are often costly and time-consuming, leading to the design of high-efficiency computational methods is in demand. In this paper, we propose a deep neural network-based method named EPIHC to predict Enhancer-Promoter Interactions with Hybrid features and Communicative learning. EPIHC extracts enhancer and promoter sequence-derived features using convolutional neural networks (CNN), and then we design a communicative learning module to capture the communicative information between enhancer and promoter sequences. Besides, EPIHC takes the genomic features of enhancers and promoters into account, incorporating with the sequence-derived features to predict EPIs. The computational experiments show that EPIHC outperforms the existing state-of-the-art EPI prediction methods on the benchmark datasets and chromosome-split datasets, and the study reveals that the communicative learning module can bring explicit information about EPIs, which is ignored by CNN, and provide explainability about EPIs to some degree. Moreover, we consider two strategies to improve the performances of EPIHC in the cross-cell line prediction, and experimental results show that EPIHC constructed on some cell lines can exhibit good performances for other cell lines. The codes and data are available at https://github.com/BioMedicalBigDataMiningLab/EPIHC.

MOTIVATION: Drug-drug interactions (DDIs) are one of the major concerns in pharmaceutical research. Many machine learning based methods have been proposed for the DDI prediction, but most of them predict whether two drugs interact or not. The studies revealed that DDIs could cause different subsequent events, and predicting DDI-associated events is more useful for investigating the mechanism hidden behind the combined drug usage or adverse reactions. RESULTS: In this article, we collect DDIs from DrugBank database, and extract 65 categories of DDI events by dependency analysis and events trimming. We propose a multimodal deep learning framework named DDIMDL that combines diverse drug features with deep learning to build a model for predicting DDI-associated events. DDIMDL first constructs deep neural network (DNN)-based sub-models, respectively, using four types of drug features: chemical substructures, targets, enzymes and pathways, and then adopts a joint DNN framework to combine the sub-models to learn cross-modality representations of drug-drug pairs and predict DDI events. In computational experiments, DDIMDL produces high-accuracy performances and has high efficiency. Moreover, DDIMDL outperforms state-of-the-art DDI event prediction methods and baseline methods. Among all the features of drugs, the chemical substructures seem to be the most informative. With the combination of substructures, targets and enzymes, DDIMDL achieves an accuracy of 0.8852 and an area under the precision-recall curve of 0.9208. AVAILABILITY AND IMPLEMENTATION: The source code and data are available at https://github.com/YifanDengWHU/DDIMDL. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

LncRNA-miRNA interactions play crucial roles in gene regulatory networks and can reveal functions of lncRNAs and miRNAs. Although several methods have been proposed to infer interactions on the lncRNA-miRNA interaction network, few attentions have been paid to fully exploiting the structure of lncRNA-miRNA interaction network. In this paper, we propose a Graph Embedding Ensemble Learning method (abbreviated as “GEEL”) to predict lncRNA-miRNA interactions. First, we collect lncRNA sequences and miRNA sequences to calculate lncRNA-lncRNA sequence similarity and miRNA-miRNA sequence similarity, and then we combine them with the known lncRNA-miRNA interactions to construct a heterogeneous network, which takes lncRNAs and miRNAs as nodes. We adopt graph embedding methods to learn representations of lncRNAs and miRNAs from the heterogeneous network, and then merge the representations of lncRNAs and miRNAs to represent the lncRNA-miRNA pairs. Random forest classifiers are built based on the merged representations to predict lncRNA-miRNA interactions. We consider five different graph embedding methods, and evaluate the corresponding models. Further, we use individual graph embedding method-based model as base predictors and build a high-level ensemble model. The experimental results show that GEEL achieves AUPR score of 0.7004 and AUC score of 0.9537, and outperforms base predictors and other state-of-the-art methods. In conclusion, GEEL is an effective tool for lncRNA-miRNA interaction prediction.

Genetic risk prediction is an important goal in human genetics research and precision medicine. Accurate prediction models will have great impacts on both disease prevention and early treatment strategies. Despite the identification of thousands of disease-associated genetic variants through genome wide association studies (GWAS), genetic risk prediction accuracy remains moderate for most diseases, which is largely due to the challenges in both identifying all the functionally relevant variants and accurately estimating their effect sizes in the presence of linkage disequilibrium. In this paper, we introduce AnnoPred, a principled framework that leverages diverse types of genomic and epigenomic functional annotations in genetic risk prediction for complex diseases. AnnoPred is trained using GWAS summary statistics in a Bayesian framework in which we explicitly model various functional annotations and allow for linkage disequilibrium estimated from reference genotype data. Compared with state-of-the-art risk prediction methods, AnnoPred achieves consistently improved prediction accuracy in both extensive simulations and real data.

Behavioral pattern discovery is increasingly being studied to understand human behavior and the discovered patterns can be used in many real world applications such as web search, recommender system and advertisement targeting. Traditional methods usually consider the behaviors as simple user and item connections, or represent them with a static model. In real world, however, human behaviors are actually complex and dynamic: they include correlations between user and multiple types of objects and also continuously evolve along time. These characteristics cause severe data sparsity and computational complexity problem, which pose great challenge to human behavioral analysis and prediction. In this paper, we propose a Flexible Evolutionary Multi-faceted Analysis (FEMA) framework for both behavior prediction and pattern mining. FEMA utilizes a flexible and dynamic factorization scheme for analyzing human behavioral data sequences, which can incorporate various knowledge embedded in different object domains to alleviate the sparsity problem. We give approximation algorithms for efficiency, where the bound of approximation loss is theoretically proved. We extensively evaluate the proposed method in two real datasets. For the prediction of human behaviors, the proposed FEMA significantly outperforms other state-of-the-art baseline methods by 17.4%. Moreover, FEMA is able to discover quite a number of interesting multi-faceted temporal patterns on human behaviors with good interpretability. More importantly, it can reduce the run time from hours to minutes, which is significant for industry to serve real-time applications.