❌

Normal view

Received — 26 March 2026 ⏭

Scaling Attention via Feature Sparsity

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yan Xie · Tiansheng Wen · Tangda Huang · Bo Chen · Chenyu You · Stefanie Jegelka · Yifei Wang
25 March 2026 at 12:00
arXiv:2603.22300v1 Announce Type: cross Abstract: Scaling Transformers to ultra-long contexts is bottlenecked by the $O(n^2 d)$ cost of self-attention. Existing methods reduce this cost along the sequence axis through local windows, kernel approximations, or token-level sparsity, but these approaches consistently degrade accuracy. In this paper, we instead explore an orthogonal axis: feature sparsity. We propose Sparse Feature Attention (SFA), where queries and keys are represented as $k$-sparse codes that preserve high-dimensional expressivity while reducing the cost of attention from $\Theta(n^2 d)$ to $\Theta(n^2 k^2/d)$. To make this efficient at scale, we introduce FlashSFA, an IO-aware kernel that extends FlashAttention to operate directly on sparse overlaps without materializing dense score matrices. Across GPT-2 and Qwen3 pretraining, SFA matches dense baselines while improving speed by up to $2.5\times$ and reducing FLOPs and KV-cache by nearly 50\%. On synthetic and downstream benchmarks, SFA preserves retrieval accuracy and robustness at long contexts, outperforming short-embedding baselines that collapse feature diversity. These results establish feature-level sparsity as a complementary and underexplored axis for efficient attention, enabling Transformers to scale to orders-of-magnitude longer contexts with minimal quality loss. Code is available at https://github.com/YannX1e/Sparse-Feature-Attention.
Received — 14 March 2026 ⏭

Targeted therapies in lung cancer: personalizing treatment across the age spectrum

✇Omics In Lung
By: Yuting Xu · Fei Chen · Honggang Zhang · Xiaofei Wang
13 March 2026 at 18:00

Front Oncol. 2026 Feb 25;16:1743620. doi: 10.3389/fonc.2026.1743620. eCollection 2026.

ABSTRACT

Lung cancer remains the leading cause of cancer-related mortality, yet current precision oncology approaches remain overwhelmingly tumor-centric, guided by genomic alterations and immune biomarkers, while largely neglecting the profound impact of aging biology on treatment response. While emerging evidence suggests that aging biology can modify therapeutic benefit and toxicity, its clinical integration remains uneven and largely investigational. In this review, we explicitly distinguish the chronological aging from biological aging to clarify how host biology modifies therapeutic benefit and toxicity. We synthesize mechanistic, translational, and early clinical evidence, while explicitly noting areas where prospective validation is lacking, to reframe personalization of lung cancer therapy through an age-conscious lens. We summarize data indicating that immunosenescence is associated with T-cell exhaustion, myeloid dominance, and extracellular matrix stiffening, features that may contribute to immune-evasive tumor phenotypes and attenuated responses to immune checkpoint blockade in subsets of patients, while pediatric cases, though rare, illustrate how global precision initiatives like iTHER and ZERO enable cautious adaptation of adult therapies. Moving beyond chronological age, we discuss biological age biomarkers, including PhenoAgeAccel, epigenetic clocks, telomere length, and frailty indices, which outperform traditional metrics in predicting risk, resistance, and toxicity, and propose integrating these tools into trial design, screening, and care planning which show promise for risk stratification and toxicity prediction but are not yet validated for routine treatment selection. Looking forward, we outline investigational strategies at the intersection of geroscience and oncology, including immune engineering, senolytics, microenvironmental modulation, and AI-driven multi-omic modeling. Overall, this review argues that biological age represents a critical but still underdeveloped dimension of precision oncology, and highlights key evidence gaps that must be addressed before age-aware personalization can be implemented in routine lung cancer care.

PMID:41821888 | PMC:PMC12975599 | DOI:10.3389/fonc.2026.1743620

Regulatory mechanisms of ALKBH5/CIITA axis in the synergistic modulation of hepatocellular carcinoma radiotherapy and immunotherapy

✇Omics in Hepatocellular
By: Fei Wang · Hui Hou · Hong Yang · Ting Dou · Zhenxia Wang · Hongdan Fu · Hong Li · Wei Zhang · Haiping Zhao · Hao Yang
11 March 2026 at 18:00

Genes Immun. 2026 Mar 10. doi: 10.1038/s41435-026-00382-6. Online ahead of print.

ABSTRACT

The prognosis for hepatocellular carcinoma remains grim. Combining radiotherapy with immune checkpoint blockade (ICB) has shown potential to enhance therapeutic outcomes, yet there is a pressing need for further advancements. Our previous research demonstrated that this combined approach suppresses ALKBH5 gene expression and increases m6A modification levels in hepatocellular carcinoma tissues. High-throughput sequencing and detailed molecular analysis revealed that inhibiting ALKBH5 amplifies CIITA m6A modifications post-therapy. This modulation triggers MHC II molecule expression in tumors, facilitating the presentation of tumor-associated antigens to CD4 + T lymphocytes and the recruitment of CD8 + T cells for an anti-tumor immune response. Building on these findings, we engineered a CIITA vector with a specific site mutation to confirm that the regulation of CIITA by the combined radiotherapy and immunotherapy is mediated through m6A methylation. Consequently, we established a comprehensive network involving ALKBH5, CIITA, MHC II, and CD4+ and CD8 + T cells. To elucidate the role and underlying molecular mechanisms of this combined therapy in reshaping the tumor immune microenvironment for hepatocellular carcinoma, we employed multi-omics approaches across in vitro, animal model, and clinical multi-dimensional studies, offering novel insights for enhancing treatment efficacy.

PMID:41807814 | DOI:10.1038/s41435-026-00382-6

Received — 5 March 2026 ⏭

T2S-Bench & Structure-of-Thought: Benchmarking and Prompting Comprehensive Text-to-Structure Reasoning

✇cs.AI, q-bio.NC updates on arXiv.org
By: Qinsi Wang · Hancheng Ye · Jinhee Kim · Jinghan Ke · Yifei Wang · Martin Kuo · Zishan Shao · Dongting Li · Yueqian Lin · Ting Jiang · Chiyue Wei · Qi Qian · Wei Wen · Helen Li · Yiran Chen
5 March 2026 at 13:00
arXiv:2603.03790v1 Announce Type: cross Abstract: Think about how human handles complex reading tasks: marking key points, inferring their relationships, and structuring information to guide understanding and responses. Likewise, can a large language model benefit from text structure to enhance text-processing performance? To explore it, in this work, we first introduce Structure of Thought (SoT), a prompting technique that explicitly guides models to construct intermediate text structures, consistently boosting performance across eight tasks and three model families. Building upon this insight, we present T2S-Bench, the first benchmark designed to evaluate and improve text-to-structure capabilities of models. T2S-Bench includes 1.8K samples across 6 scientific domains and 32 structural types, rigorously constructed to ensure accuracy, fairness, and quality. Evaluation on 45 mainstream models reveals substantial improvement potential: the average accuracy on the multi-hop reasoning task is only 52.1%, and even the most advanced model achieves 58.1% node accuracy in end-to-end extraction. Furthermore, on Qwen2.5-7B-Instruct, SoT alone yields an average +5.7% improvement across eight diverse text-processing tasks, and fine-tuning on T2S-Bench further increases this gain to +8.6%. These results highlight the value of explicit text structuring and the complementary contributions of SoT and T2S-Bench. Dataset and eval code have been released at https://t2s-bench.github.io/T2S-Bench-Page/.

IPD: Boosting Sequential Policy with Imaginary Planning Distillation in Offline Reinforcement Learning

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yihao Qin · Yuanfei Wang · Hang Zhou · Peiran Liu · Hao Dong · Yiding Ji
5 March 2026 at 13:00
arXiv:2603.04289v1 Announce Type: cross Abstract: Decision transformer based sequential policies have emerged as a powerful paradigm in offline reinforcement learning (RL), yet their efficacy remains constrained by the quality of static datasets and inherent architectural limitations. Specifically, these models often struggle to effectively integrate suboptimal experiences and fail to explicitly plan for an optimal policy. To bridge this gap, we propose \textbf{Imaginary Planning Distillation (IPD)}, a novel framework that seamlessly incorporates offline planning into data generation, supervised training, and online inference. Our framework first learns a world model equipped with uncertainty measures and a quasi-optimal value function from the offline data. These components are utilized to identify suboptimal trajectories and augment them with reliable, imagined optimal rollouts generated via Model Predictive Control (MPC). A Transformer-based sequential policy is then trained on this enriched dataset, complemented by a value-guided objective that promotes the distillation of the optimal policy. By replacing the conventional, manually-tuned return-to-go with the learned quasi-optimal value function, IPD improves both decision-making stability and performance during inference. Empirical evaluations on the D4RL benchmark demonstrate that IPD significantly outperforms several state-of-the-art value-based and transformer-based offline RL methods across diverse tasks.

MIRAGE: Knowledge Graph-Guided Cross-Cohort MRI Synthesis for Alzheimer's Disease Prediction

✇cs.AI, q-bio.NC updates on arXiv.org
By: Guanchen Wu · Zhe Huang · Yuzhang Xie · Runze Yan · Akul Chopra · Deqiang Qiu · Xiao Hu · Fei Wang · Carl Yang
4 March 2026 at 13:00
arXiv:2603.02434v1 Announce Type: cross Abstract: Reliable Alzheimer's disease (AD) diagnosis increasingly relies on multimodal assessments combining structural Magnetic Resonance Imaging (MRI) and Electronic Health Records (EHR). However, deploying these models is bottlenecked by modality missingness, as MRI scans are expensive and frequently unavailable in many patient cohorts. Furthermore, synthesizing de novo 3D anatomical scans from sparse, high-dimensional tabular records is technically challenging and poses severe clinical risks. To address this, we introduce MIRAGE, a novel framework that reframes the missing-MRI problem as an anatomy-guided cross-modal latent distillation task. First, MIRAGE leverages a Biomedical Knowledge Graph (KG) and Graph Attention Networks to map heterogeneous EHR variables into a unified embedding space that can be propagated from cohorts with real MRIs to cohorts without them. To bridge the semantic gap and enforce physical spatial awareness, we employ a frozen pre-trained 3D U-Net decoder strictly as an auxiliary regularization engine. Supported by a novel cohort-aggregated skip feature compensation strategy, this decoder acts as a rigorous structural penalty, forcing 1D latent representations to encode biologically plausible, macro-level pathological semantics. By exclusively utilizing this distilled "diagnostic-surrogate" representation during inference, MIRAGE completely bypasses computationally expensive 3D voxel reconstruction. Experiments demonstrate that our framework successfully bridges the missing-modality gap, improving the AD classification rate by 13% compared to unimodal baselines in cohorts without real MRIs.

RubricBench: Aligning Model-Generated Rubrics with Human Standards

✇cs.AI, q-bio.NC updates on arXiv.org
By: Qiyuan Zhang · Junyi Zhou · Yufei Wang · Fuyuan Lyu · Yidong Ming · Can Xu · Qingfeng Sun · Kai Zheng · Peng Kang · Xue Liu · Chen Ma
4 March 2026 at 13:00
arXiv:2603.01562v2 Announce Type: replace Abstract: As Large Language Model (LLM) alignment evolves from simple completions to complex, highly sophisticated generation, Reward Models are increasingly shifting toward rubric-guided evaluation to mitigate surface-level biases. However, the community lacks a unified benchmark to assess this evaluation paradigm, as existing benchmarks lack both the discriminative complexity and the ground-truth rubric annotations required for rigorous analysis. To bridge this gap, we introduce RubricBench, a curated benchmark with 1,147 pairwise comparisons specifically designed to assess the reliability of rubric-based evaluation. Our construction employs a multi-dimensional filtration pipeline to target hard samples featuring nuanced input complexity and misleading surface bias, augmenting each with expert-annotated, atomic rubrics derived strictly from instructions. Comprehensive experiments reveal a substantial capability gap between human-annotated and model-generated rubrics, indicating that even state-of-the-art models struggle to autonomously specify valid evaluation criteria, lagging considerably behind human-guided performance.
Received — 17 February 2026 ⏭

Transferable XAI: Relating Understanding Across Domains with Explanation Transfer

✇cs.AI, q-bio.NC updates on arXiv.org
By: Fei Wang · Yifan Zhang · Brian Y. Lim
17 February 2026 at 13:00
arXiv:2602.13675v1 Announce Type: cross Abstract: Current Explainable AI (XAI) focuses on explaining a single application, but when encountering related applications, users may rely on their prior understanding from previous explanations. This leads to either overgeneralization and AI overreliance, or burdensome independent memorization. Indeed, related decision tasks can share explanatory factors, but with some notable differences; e.g., body mass index (BMI) affects the risks for heart disease and diabetes at the same rate, but chest pain is more indicative of heart disease. Similarly, models using different attributes for the same task still share signals; e.g., temperature and pressure affect air pollution but in opposite directions due to the ideal gas law. Leveraging transfer of learning, we propose Transferable XAI to enable users to transfer understanding across related domains by explaining the relationship between domain explanations using a general affine transformation framework applied to linear factor explanations. The framework supports explanation transfer across various domain types: translation for data subspace (subsuming prior work on Incremental XAI), scaling for decision task, and mapping for attributes. Focusing on task and attributes domain types, in formative and summative user studies, we investigated how well participants could understand AI decisions from one domain to another. Compared to single-domain and domain-independent explanations, Transferable XAI was the most helpful for understanding the second domain, leading to the best decision faithfulness, factor recall, and ability to relate explanations between domains. This framework contributes to improving the reusability of explanations across related AI applications by explaining factor relationships between subspaces, tasks, and attributes.

Comparables XAI: Faithful Example-based AI Explanations with Counterfactual Trace Adjustments

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yifan Zhang · Tianle Ren · Fei Wang · Brian Y Lim
17 February 2026 at 13:00
arXiv:2602.13784v1 Announce Type: cross Abstract: Explaining with examples is an intuitive way to justify AI decisions. However, it is challenging to understand how a decision value should change relative to the examples with many features differing by large amounts. We draw from real estate valuation that uses Comparables-examples with known values for comparison. Estimates are made more accurate by hypothetically adjusting the attributes of each Comparable and correspondingly changing the value based on factors. We propose Comparables XAI for relatable example-based explanations of AI with Trace adjustments that trace counterfactual changes from each Comparable to the Subject, one attribute at a time, monotonically along the AI feature space. In modelling and user studies, Trace-adjusted Comparables achieved the highest XAI faithfulness and precision, user accuracy, and narrowest uncertainty bounds compared to linear regression, linearly adjusted Comparables, or unadjusted Comparables. This work contributes a new analytical basis for using example-based explanations to improve user understanding of AI decisions.
There are no more articles
❌